Sowter: Significantly less inductance than originals with respect to their RCA BA-6 and 86 interstage offerings. Much lower DC resistance, larger wire, less turns. I have two reports now of unsatisfactory performance of the 86 type when used to replace an original transformer on a real RCA 86. One report was -7 at 100Hz, the other -30 at 100Hz. We have people building DRIP BA-6's don't we? I have yet to see real response and distortion measurements from a clone build.
RCA/175 Style Limiter DIY
- Thread starter alexc
- Start date
Help Support GroupDIY Audio Forum:
Ouch! That would hurt - not an insignificant investment and to have it perform below expectations must be a real slam.
I guess one needs to check explicitly and not presume any particular manufacturer.
I hope they can improve the ba6a style units - I'm not sure I want to live in a world where Sowter cant make the best traffo out there :'(
However, I feel a little better - I'm not alone in being challenged with low freq extension here
Would be great to hear some originals or working copies. Although, as you mentioned in the case of the original 175, even that's not necessarily brilliant.
Oh well, back to the drawing board.
I guess one needs to check explicitly and not presume any particular manufacturer.
I hope they can improve the ba6a style units - I'm not sure I want to live in a world where Sowter cant make the best traffo out there :'(
However, I feel a little better - I'm not alone in being challenged with low freq extension here
Would be great to hear some originals or working copies. Although, as you mentioned in the case of the original 175, even that's not necessarily brilliant.
Oh well, back to the drawing board.
So now working backwards, from the finals forward to see what's what with low freq.
OK - with audio directly applied to the finals grids, via 1uF cap no problem.
20Hz is virtually flat at -0.5dB, 20KHz, +0.3dB
Backing up the circuit to the 0.01uF coupling caps on the grids of the finals.
Oh! That's not good. They go now.
With the au7 driver in circuit coupled with 1uF, completely flat.
Bypassing the interstage traffo with cap coupling 4.7uF instead, totally flat.
Now with the pentode GR stages in circuit and restoring the interstage traffo
----
So increasing the last coupling cap from 0.01uF extends the lf extension.
Here's a shot as it stands with 1uF and the interstage - looking like 55Hz or so.
It's a little jagged because I'm running the interstage with high gain - there's a bit more noise in there.
0.1uF is about right, so I'm going with that for now.
Just for checks, I put the original interstage back in - it remains pretty bad
Right, now to button that up and return to gain staging thinks.
OK - with audio directly applied to the finals grids, via 1uF cap no problem.
20Hz is virtually flat at -0.5dB, 20KHz, +0.3dB
Backing up the circuit to the 0.01uF coupling caps on the grids of the finals.
Oh! That's not good. They go now.
With the au7 driver in circuit coupled with 1uF, completely flat.
Bypassing the interstage traffo with cap coupling 4.7uF instead, totally flat.
Now with the pentode GR stages in circuit and restoring the interstage traffo
----
So increasing the last coupling cap from 0.01uF extends the lf extension.
Here's a shot as it stands with 1uF and the interstage - looking like 55Hz or so.
It's a little jagged because I'm running the interstage with high gain - there's a bit more noise in there.
0.1uF is about right, so I'm going with that for now.
Just for checks, I put the original interstage back in - it remains pretty bad
Right, now to button that up and return to gain staging thinks.
Attachments
Because I have a low ratio step up at the input transformer now (5K:10K), my gain staging is lower overall - too low! , so I am setting the interstage up for 2K:10K which is x2.23 step up.
----
Now when I re-introduce the 5K:10K input traffo, it dominates the low freq response, bringing it back to 123HZ or so.
Testing with wsm 10K:10K as input traffo shows similar.
Testing with the xsm 600:10K does the trick - flat all the way up and -3dB at 36Hz or so.
It's a bit jaggy because I have a tonne of gain now but that's pretty clear.
And again with a stancor 1K:10K or so very similar as xsm.
And again with a generic 600:600 has the big rolloff and peaking at hi.
Now this is all with :
xlr -> input traffo -> balanced rotary atten -> GR pentodes -> interstage -> line amp
So - need to find a new pair with a decent step up 600:10K or so.
That's back where I started and I can set the interstage to a modest step down again.
----
Now just to labour the point, I'm using an a15 500:10K as input and another a15 as interstage 10K:2K stepdown. Let's see what that does!
Not so good - 100Hz
----
So far best is edcor xsm 600:10K input and a15 10K:2K interstage and 0.1uF coupling cap to finals.
Both sides show same result more or less - the octal side has a bit more rolloff - I think that one could benefit with a 0.22uF or higher at the coupling caps.
Not sure where to go - don't want to use xsm here - no new buys of them for me.
On the other hand, this is definately a try-before-buy build .. so I may need to try a few from other builds till I get something I know works .. then I can buy .. after I sell some stuff to pay, that is :-X
-------
a-series are out - too susceptible to noise, not going thru all that again!
I have a high step up ouncer to try for the hell of it and I have a sowter step up from an la2a to try, as well as some jensen 2.5K:40K (1:4) or so from another la2a
My guess is the jensens will work best - large in size and good step up, screening and full metal jacket , far low end extension flat well beyond 20K.
I can sub for the la2a build alot easier than this difficult beast - you can stick just about anything in an la2a and they stay great
In the interests of science, I'll mess with those completed and very well working builds to find a path forward!
----
Now for some deep thought ...
----
Now when I re-introduce the 5K:10K input traffo, it dominates the low freq response, bringing it back to 123HZ or so.
Testing with wsm 10K:10K as input traffo shows similar.
Testing with the xsm 600:10K does the trick - flat all the way up and -3dB at 36Hz or so.
It's a bit jaggy because I have a tonne of gain now but that's pretty clear.
And again with a stancor 1K:10K or so very similar as xsm.
And again with a generic 600:600 has the big rolloff and peaking at hi.
Now this is all with :
xlr -> input traffo -> balanced rotary atten -> GR pentodes -> interstage -> line amp
So - need to find a new pair
with a decent step up 600:10K or so. That's back where I started and I can set the interstage to a modest step down again.
----
Now just to labour the point, I'm using an a15 500:10K as input and another a15 as interstage 10K:2K stepdown.
Let's see what that does!Not so good - 100Hz
----
So far best is edcor xsm 600:10K input and a15 10K:2K interstage and 0.1uF coupling cap to finals.
Both sides show same result more or less - the octal side has a bit more rolloff - I think that one could benefit with a 0.22uF or higher at the coupling caps.
Not sure where to go - don't want to use xsm here - no new buys of them for me.
On the other hand, this is definately a try-before-buy build .. so I may need to try a few from other builds till I get something I know works .. then I can buy .. after I sell some stuff to pay, that is :-X
-------
a-series are out - too susceptible to noise, not going thru all that again!
I have a high step up ouncer to try for the hell of it and I have a sowter step up from an la2a to try, as well as some jensen 2.5K:40K (1:4) or so from another la2a
My guess is the jensens will work best - large in size and good step up, screening and full metal jacket , far low end extension flat well beyond 20K.
I can sub for the la2a build alot easier than this difficult beast - you can stick just about anything in an la2a and they stay great
In the interests of science, I'll mess with those completed and very well working builds to find a path forward!
----
Now for some deep thought ...
Just backtracking now to re-establish my baselines ..
- removed input transformer
- interstage transformer is in circuit with 10K:2K (2.24:1) step down
- both channels same except coupling caps (and tube sets)
The noval channel is very good and flat with -3dB at 40Hz with 0.1uF coupling caps at the finals
The octal channel is worse and flat with -3dB at 80Hz with 1uF coupling caps at the finals
So trying some things to get the octals the same.
---
Now this is all quite level dependent - so I've subbed the au7 for at7 and have the gains across both channels pretty much the same.
(the 6V6 final is somewhat higher gain than 6S4A - more current to the pp output traffo)
So now I have -3dB low freq points
noval 50Hz and octal 80Hz (octal is -1.5dB worse at 50Hz)
Unity gains noisefloors at these gain staging are (20-20KHz, 'spl style' measurement)
octal -83.3dBu, noval -76.0dBu (closest to psu)
Overall gain (input at minimum attenuation) from xlr in to output attenuator input is around x21 or +26dB for each. The output atten is almost at max atten at 7 o'clock (or notch 1 from 10)
No feedback.
----
I played about with cathode bypass here and there with no improvement in low freq extension after compensating for the gain changes which resulted.
Anyway, getting closer to my next baseline.
Next thing I'm going to test is driving this unit 'transformeless' from the brown preamp, which has xsm 15K:600 outs and loads of push.
I want to see if can drive without messing things up (more)
- removed input transformer
- interstage transformer is in circuit with 10K:2K (2.24:1) step down
- both channels same except coupling caps (and tube sets)
The noval channel is very good and flat with -3dB at 40Hz with 0.1uF coupling caps at the finals
The octal channel is worse and flat with -3dB at 80Hz with 1uF coupling caps at the finals
So trying some things to get the octals the same.
---
Now this is all quite level dependent - so I've subbed the au7 for at7 and have the gains across both channels pretty much the same.
(the 6V6 final is somewhat higher gain than 6S4A - more current to the pp output traffo)
So now I have -3dB low freq points
noval 50Hz and octal 80Hz (octal is -1.5dB worse at 50Hz)
Unity gains noisefloors at these gain staging are (20-20KHz, 'spl style' measurement)
octal -83.3dBu, noval -76.0dBu (closest to psu)
Overall gain (input at minimum attenuation) from xlr in to output attenuator input is around x21 or +26dB for each. The output atten is almost at max atten at 7 o'clock (or notch 1 from 10)
No feedback.
----
I played about with cathode bypass here and there with no improvement in low freq extension after compensating for the gain changes which resulted.
Anyway, getting closer to my next baseline.
Next thing I'm going to test is driving this unit 'transformeless' from the brown preamp, which has xsm 15K:600 outs and loads of push.
I want to see if can drive without messing things up (more)
Attachments
Good news everybody!
That tested up really well - the 'Brown Pre' as I call it, has Chn A as a Blonde Bassman (6G6) and the Chn B as a Brown Twin Amp with Harmonic Vibrato (6G8).
Now the Bassman eq can go virtually flat, something like -3dB 20Hz, -3dB 20KHz. There's a very slight rise at around 100Hz and also at around 5KHz.
So it is the 'hifi' channel, with a fully extended bottom end, well suited to bass.
The Vibro is set to have a quite sharp low freq rolloff with -3dB point at around 90Hz and also has a gentle Hi freq rolloff with -3dB at 10KHz or so. Within the passband, it can be eq'd pretty flat.
I set up that passband because the vibro needs steep HPF to prevent trem osc punch-thru at strong vibro settings, as well because it has a switchable additional triode 'crunch' stage.
(trem punch-thru has a lot of parallels to CV punch-thru or 'thumping' )
The LPF at the top end helps avoid fizz/ice-pick in either the trem and/or gain settings.
---
OK so, I set up both the Brown channels for flat, and same gain as my previous tests : -10dBu at the XLR.
That's hugely throttled! The Brown can easily do 50Vpp from it's finals with B+ 270V white cathode follower 15K:600 output.
Into the dual RCA style box and using the RTAS to send/receive balanced test signal as usual.
Looking at the freq response of all that - no problemo.
Bassman channel is a good match for the RCA noval channel, with it's 50Hz low freq extension
Vibro channel doesn't suffer on the RCA octal channel, with it's 80Hz low freq point.
The upshot is, everything looks pretty much the same as the RCA on it's own, freq response wise.
Also, the balance at the output of the RCA's 25K dual rotary attenuators (post XLR and pre GR pentodes) was fine - each phase nice and equal.
So, looks like I can run the RCA without an input transformer. The Brown pre has no problem whatever driving the RCA and it's high output is an advantage in this 'transformerless' input scenario.
-----
NOTE
With the CV dc voltage fed to the CT of the dual rotary, it will be present at the XLR input pins (+) and (-). In my case, I am feeding this with a floating transformer output from the driving device, so there is not a problem. SO - this configuration is ONLY suitable for driving with a hefty floating traffo - anything else would see a sizeable -ve dc offset which could FRY a SS output stage!
It could also be a problem with the Edcor XSM at the output of the driving equipment - I'll just have to see. Perhaps there will be enough imbalance in currents due to the offset so as to cause loss of inductance, but based on the usage in the pm660, I don' think it should be an issue.
You could, of course, add some caps at the xlr ins. Not me, though.
------
Typically, I can send out 12Vpp easy, have -6dB on the attenuator and still have 6Vpp at the grids of the GR tubes, which are biased to sit around +5Vdc or so, wanting 10Vpp max at idle.
The Brown pre can easily do a whole lot more and indeed it's difficult to hold it back. Currently it feeds my Motu which bleats like a lonely sheep at 9 - 10Vpp max.
My guitars, which typically use active pickups or stacked noiseless passives, have a large clean dynamic range - typically 2Vpp and reaching 4Vpp on hard transients.
The Brown pre, with grids at around 2.75V or so, excels with all that, but stresses out the girly Motu inputs quite a bit.
Especially the bass. A Rick bass with actives into a tube pre is something to be wary of!
I suspect there is a poor impedance match there too because the transients seem to overshoot and clip quite a bit more than comparable level steady-state material do.
(I terminate in 600R 2W resistors internally)
----
SO - the 'transformerless' mode of the RCA limiter is a brilliant match for the Brown pre and eliminates the usual large 'step up' and attendant large 'attenuate down' before hitting the GR grids.
Without, seemingly, screwing up freq resp and what not.
I have obviously yet to do detail combined noise floor tests but I did a quick play test and it sounded fine, with plenty gain in hand and not seeing/hearing any issues at all.
I know the sound of my Jazzmaster very well - I've had it for 25years or so and it was always my main git when I used to play the clubs as a young EE student and 'cadet' engineer.
So - I'm quite happy about that ;D
----
ps - the brown pre wires are all about switching of ins/outs and also direct hiZ outs from cathode circuits. Another 'prototype'
pps - Sorry about the fuzzy pic!
ppps - I use ay7 for signal, bh7 for line amp and ax7 for trem.
That tested up really well - the 'Brown Pre' as I call it, has Chn A as a Blonde Bassman (6G6) and the Chn B as a Brown Twin Amp with Harmonic Vibrato (6G8).
Now the Bassman eq can go virtually flat, something like -3dB 20Hz, -3dB 20KHz. There's a very slight rise at around 100Hz and also at around 5KHz.
So it is the 'hifi' channel, with a fully extended bottom end, well suited to bass.
The Vibro is set to have a quite sharp low freq rolloff with -3dB point at around 90Hz and also has a gentle Hi freq rolloff with -3dB at 10KHz or so. Within the passband, it can be eq'd pretty flat.
I set up that passband because the vibro needs steep HPF to prevent trem osc punch-thru at strong vibro settings, as well because it has a switchable additional triode 'crunch' stage.
(trem punch-thru has a lot of parallels to CV punch-thru or 'thumping'
)The LPF at the top end helps avoid fizz/ice-pick in either the trem and/or gain settings.
---
OK so, I set up both the Brown channels for flat, and same gain as my previous tests : -10dBu at the XLR.
That's hugely throttled! The Brown can easily do 50Vpp from it's finals with B+ 270V white cathode follower 15K:600 output.
Into the dual RCA style box and using the RTAS to send/receive balanced test signal as usual.
Looking at the freq response of all that - no problemo.
Bassman channel is a good match for the RCA noval channel, with it's 50Hz low freq extension
Vibro channel doesn't suffer on the RCA octal channel, with it's 80Hz low freq point.
The upshot is, everything looks pretty much the same as the RCA on it's own, freq response wise.
Also, the balance at the output of the RCA's 25K dual rotary attenuators (post XLR and pre GR pentodes) was fine - each phase nice and equal.
So, looks like I can run the RCA without an input transformer. The Brown pre has no problem whatever driving the RCA and it's high output is an advantage in this 'transformerless' input scenario. -----
NOTE
With the CV dc voltage fed to the CT of the dual rotary, it will be present at the XLR input pins (+) and (-). In my case, I am feeding this with a floating transformer output from the driving device, so there is not a problem. SO - this configuration is ONLY suitable for driving with a hefty floating traffo - anything else would see a sizeable -ve dc offset which could FRY a SS output stage!
It could also be a problem with the Edcor XSM at the output of the driving equipment - I'll just have to see. Perhaps there will be enough imbalance in currents due to the offset so as to cause loss of inductance, but based on the usage in the pm660, I don' think it should be an issue.
You could, of course, add some caps at the xlr ins. Not me, though.
------
Typically, I can send out 12Vpp easy, have -6dB on the attenuator and still have 6Vpp at the grids of the GR tubes, which are biased to sit around +5Vdc or so, wanting 10Vpp max at idle.
The Brown pre can easily do a whole lot more and indeed it's difficult to hold it back. Currently it feeds my Motu which bleats like a lonely sheep at 9 - 10Vpp max.
My guitars, which typically use active pickups or stacked noiseless passives, have a large clean dynamic range - typically 2Vpp and reaching 4Vpp on hard transients.
The Brown pre, with grids at around 2.75V or so, excels with all that, but stresses out the girly Motu inputs quite a bit.
Especially the bass. A Rick bass with actives into a tube pre is something to be wary of!
I suspect there is a poor impedance match there too because the transients seem to overshoot and clip quite a bit more than comparable level steady-state material do.
(I terminate in 600R 2W resistors internally)
----
SO - the 'transformerless' mode of the RCA limiter is a brilliant match for the Brown pre and eliminates the usual large 'step up' and attendant large 'attenuate down' before hitting the GR grids.
Without, seemingly, screwing up freq resp and what not.
I have obviously yet to do detail combined noise floor tests but I did a quick play test and it sounded fine, with plenty gain in hand and not seeing/hearing any issues at all.
I know the sound of my Jazzmaster very well - I've had it for 25years or so and it was always my main git when I used to play the clubs as a young EE student and 'cadet' engineer.
So - I'm quite happy about that ;D
----
ps - the brown pre wires are all about switching of ins/outs and also direct hiZ outs from cathode circuits. Another 'prototype'
pps - Sorry about the fuzzy pic!
ppps - I use ay7 for signal, bh7 for line amp and ax7 for trem.
Attachments
I will persevere and try to get the octal and noval as similar as I can - but at least I have a working scenario right now.
So - continued futzing about to improve the noise floor of the noval and freq response of the octal.
But - I'll take a break from input traffo issues for a bit while I move forward some.
As well - I'm be wiring up the attack/release and get the sidechain working right. The dc is already done.
Finally - I will further consider the gain staging : with a low-ish +26dB (excluding attenuators), I should be OK for 15dB GR BUT may well end up needing some more in hand for feedback (maybe) and more attenuator control (possibly).
I could wind up with 6SL7 in place of 6SN7 and and an AX7 in place of AT7 if I am forced to
So - some compression soon
Goodnight everybody ;D
So - continued futzing about to improve the noise floor of the noval and freq response of the octal.
But - I'll take a break from input traffo issues for a bit while I move forward some.
As well - I'm be wiring up the attack/release and get the sidechain working right. The dc is already done.
Finally - I will further consider the gain staging : with a low-ish +26dB (excluding attenuators), I should be OK for 15dB GR BUT may well end up needing some more in hand for feedback (maybe) and more attenuator control (possibly).
I could wind up with 6SL7 in place of 6SN7 and and an AX7 in place of AT7 if I am forced to
So - some compression soon
Goodnight everybody ;D
Attachments
Curiouser and curiouser ..
I removed those mystery utc LH-519 and did some basic tests on the bench.
Using a simple freq response test, driven directly from my Motu interface outputs and connected directly back to the Motu inputs :
completely and totally flat 10Hz - 20KHz
OK - with full zoom in, at +4dBu excitation, there is a -1.3dB wobbly at 50Hz - the rest is flat.
That's the worst case.
Really - all the ratios check out as expected, no difference which way 10K:5K or 5K:10K and even using the CT on each side. All flat.
I tried loading each side resistively - no difference. Flat.
I tried different amplitude sweeps, low to quite high. Again flat.
Normally I see the 'small' frame ones with suspected low inductance show a significant rolloff, even in this test. Most of the larger ones, with supposed higher inductance don't even go flat to 10Hz.
-----
OK - so there it is. I'm certainly able to use these in a full bandwidth build - just not this RCA style one. Apparently.
I removed those mystery utc LH-519 and did some basic tests on the bench.
Using a simple freq response test, driven directly from my Motu interface outputs and connected directly back to the Motu inputs :
completely and totally flat 10Hz - 20KHz
OK - with full zoom in, at +4dBu excitation, there is a -1.3dB wobbly at 50Hz - the rest is flat.
That's the worst case.
Really - all the ratios check out as expected, no difference which way 10K:5K or 5K:10K and even using the CT on each side. All flat.
I tried loading each side resistively - no difference. Flat.
I tried different amplitude sweeps, low to quite high. Again flat.
Normally I see the 'small' frame ones with suspected low inductance show a significant rolloff, even in this test. Most of the larger ones, with supposed higher inductance don't even go flat to 10Hz.
-----
OK - so there it is. I'm certainly able to use these in a full bandwidth build - just not this RCA style one. Apparently.
Attachments
Now the 'Whitlock whitepaper' has the diagram shown below which gives the basic mechanism for the freq response of the transformer.
As I said before, Rg is unchanged between the in-circuit and out-of-circuit tests.
So it must be the different RL which is worsening the low-freq in-circuit.
ie. the 25K balanced rotary with ground referenced centre junction
(the utc LH519 has single coils each side with a centre-tap - I did not ground the coil CT, only grounded the junction of the 2 atten decks)
So the RL would be that attenuator in parallel with the grid of the pentode (6SS7 octals, 6BJ6 novals).
In the RCA circuit, there are no additional 'grid leak' resistors.
RL should be relatively small; 25K max and with certain attenuator settings, even less.
Maybe the pentode plate-grid (miller) cap ? I thought that was quite small in pentodes due to the inluence of the screen.
----
So wtf is going on ? Why would a flat response input traffo cause a 75Hz shift to the right with the same driving impedance and a lowish load impedance ?
I'll do more tests in-circuit later ..... For now, I want to press on.
As I said before, Rg is unchanged between the in-circuit and out-of-circuit tests.
So it must be the different RL which is worsening the low-freq in-circuit.
ie. the 25K balanced rotary with ground referenced centre junction
(the utc LH519 has single coils each side with a centre-tap - I did not ground the coil CT, only grounded the junction of the 2 atten decks)
So the RL would be that attenuator in parallel with the grid of the pentode (6SS7 octals, 6BJ6 novals).
In the RCA circuit, there are no additional 'grid leak' resistors.
RL should be relatively small; 25K max and with certain attenuator settings, even less.
Maybe the pentode plate-grid (miller) cap ? I thought that was quite small in pentodes due to the inluence of the screen.
----
So wtf is going on ? Why would a flat response input traffo cause a 75Hz shift to the right with the same driving impedance and a lowish load impedance ?
I'll do more tests in-circuit later ..... For now, I want to press on.
Attachments
The other thing I don't follow is why I have a 30Hz worsening of the low freq response (no input traffo) between the octal and noval cases.
As I said, cathode coupling cap changes at the 6SN7 driver makes no change to the low freq extension, nor at the 6V6.
Everything else is the same, except tube sets. Gains are very close across each side.
Increasing the octal finals' coupling cap to 1uF from 0.1uF (as novals have) didn't make much difference either, although I left those in there.
The 'loss of low freq' happens at the interstage traffo ie. when the 6sn7 is directly driven from Motu, the low freq is much extended.
In the noval AT7 case, that difference made by interstage traffo is a lot smaller.
So - something related to how the 6SN7 grid circuit (as compared to the AT7) is loading the secondary of the interstage .. higher?
ie. RL of the insterstage increased so shifting to the right
The driver grid circuit is the same in both sides : a balanced pair of 22K, center junction to ground, similar to the ua175b case.
I want answers
As I said, cathode coupling cap changes at the 6SN7 driver makes no change to the low freq extension, nor at the 6V6.
Everything else is the same, except tube sets. Gains are very close across each side.
Increasing the octal finals' coupling cap to 1uF from 0.1uF (as novals have) didn't make much difference either, although I left those in there.
The 'loss of low freq' happens at the interstage traffo ie. when the 6sn7 is directly driven from Motu, the low freq is much extended.
In the noval AT7 case, that difference made by interstage traffo is a lot smaller.
So - something related to how the 6SN7 grid circuit (as compared to the AT7) is loading the secondary of the interstage .. higher?
ie. RL of the insterstage increased so shifting to the right
The driver grid circuit is the same in both sides : a balanced pair of 22K, center junction to ground, similar to the ua175b case.
I want answers
Forging on with the meters.
I have wired up my GR/VU switch for the meter - they are MCI/Triplett units with MCI meter buffer pcb. That's a nice little circuit - a single 741 opamp (yep!) in differential mode with bipolar supply.
It has a trimmer in one input arm for adjust and drives the meter in single-ended fashion.
This build uses the same psu traffo as the PM660, so it has a utility CT winding 15-0-15ac - making it pretty easy for me. I will do a simple rectifier+cap smoothing board. No regulation needed here to drive the utilities.
So I hacked into the pcb so I could switch out the buffer when using the 'GR' mode. The meter is a FW diode bridge rectified type (germanium I think).
----
OK - so I have a simplified GR feed to the meter - even more simple than the ua175b. It is fed from the cathode circuit and goes to a 1K pot in parallel with the meter.
Yay - it works and they zero up nicely and is stable and not drifitng around.
It's not as fine a 'zero' as the ua175b build because I don't have that somewhat groovier resistive network around the zero pot that the ua175 has.
I look at that more when I have liftoff with GR.
-----
I have the attack, release pots done - same as the ua175b. 25K attack pot, 5M release pot, 0.1uF wima as the release cap.
I added a threshhold pot, which varies the dc bias to the cathodes of the 6AL5 rectifier.
This is similar to the DaveP RCA BA-25 but using a 6AL5 rectifier.
At present does 0-100V of bias variation - I may adjust that once I get going with GR.
The Drip 175b build was a fixed +40V dc offset to the plates of the 6AL5.
Right - now to test the GR
I have wired up my GR/VU switch for the meter - they are MCI/Triplett units with MCI meter buffer pcb. That's a nice little circuit - a single 741 opamp (yep!) in differential mode with bipolar supply.
It has a trimmer in one input arm for adjust and drives the meter in single-ended fashion.
This build uses the same psu traffo as the PM660, so it has a utility CT winding 15-0-15ac - making it pretty easy for me. I will do a simple rectifier+cap smoothing board. No regulation needed here to drive the utilities.
So I hacked into the pcb so I could switch out the buffer when using the 'GR' mode. The meter is a FW diode bridge rectified type (germanium I think).
----
OK - so I have a simplified GR feed to the meter - even more simple than the ua175b. It is fed from the cathode circuit and goes to a 1K pot in parallel with the meter.
Yay - it works and they zero up nicely and is stable and not drifitng around.
It's not as fine a 'zero' as the ua175b build because I don't have that somewhat groovier resistive network around the zero pot that the ua175 has.
I look at that more when I have liftoff with GR.
-----
I have the attack, release pots done - same as the ua175b. 25K attack pot, 5M release pot, 0.1uF wima as the release cap.
I added a threshhold pot, which varies the dc bias to the cathodes of the 6AL5 rectifier.
This is similar to the DaveP RCA BA-25 but using a 6AL5 rectifier.
At present does 0-100V of bias variation - I may adjust that once I get going with GR.
The Drip 175b build was a fixed +40V dc offset to the plates of the 6AL5.
Right - now to test the GR
Well the Threshold is all working right as is the rectifier, attack and release.
I have -10dBu 1KHz as input fed thru an isolation transformer
*** ISO is REQUIRED to prevent CV at the XLR input pins from getting to SS output stages of my Motu ***
Input Atten set to minimum attenuation and I can easily see CV of 0 .. -11V or so depending on the Threshhold setting - CV is killed fully at around +85V threshold or so
Thats around 8dB of GR on the noval channel (6BJ6 at present - will soon goto 6BD8)
and around 7dB of GR on the octal channel (6SS7).
The GR meters are showing more like 15dB of GR in each case, so further work required there.
-----
Now I'm barely tickling this guy right now, with a very underwhelming -10dBu input. So 7dB GR (with CV of -11V for a total grid-cathode diff of around -16V) right now is pretty good.
I expect that when I have something more like -20V of CV I will begin to approach 'cut off' and the GR will shoot right up.
If necessary I can further fine tune with the screen supply - I have a fairly modest +100V right now.
So having the Threshhold to fine tune the CV at whatever input you want is not only a 'civilized' touch but imho absolutely necessary. Once you have it you never go back!
Anything else is a little neanderthal (no offense to you neanderthals out there)
----
I'll probably change the 25K attack pots to switched pot versions so as to have a 'soft bypass' in addition to the 'relay hard bypass' that I am shortly to add - you can do it with the Threshhold control so for now no big deal.
So the basic thing is working 8) 8)
----
For reference, here's my Motu 828 MkII -10dBu 1KHz sine loopback spectra
I have -10dBu 1KHz as input fed thru an isolation transformer
*** ISO is REQUIRED to prevent CV at the XLR input pins from getting to SS output stages of my Motu ***
Input Atten set to minimum attenuation and I can easily see CV of 0 .. -11V or so depending on the Threshhold setting - CV is killed fully at around +85V threshold or so
Thats around 8dB of GR on the noval channel (6BJ6 at present - will soon goto 6BD8)
and around 7dB of GR on the octal channel (6SS7).
The GR meters are showing more like 15dB of GR in each case, so further work required there.
-----
Now I'm barely tickling this guy right now, with a very underwhelming -10dBu input. So 7dB GR (with CV of -11V for a total grid-cathode diff of around -16V) right now is pretty good.
I expect that when I have something more like -20V of CV I will begin to approach 'cut off' and the GR will shoot right up.
If necessary I can further fine tune with the screen supply - I have a fairly modest +100V right now.
So having the Threshhold to fine tune the CV at whatever input you want is not only a 'civilized' touch but imho absolutely necessary. Once you have it you never go back!
Anything else is a little neanderthal (no offense to you neanderthals out there)
----
I'll probably change the 25K attack pots to switched pot versions so as to have a 'soft bypass' in addition to the 'relay hard bypass' that I am shortly to add - you can do it with the Threshhold control so for now no big deal.
So the basic thing is working 8) 8)
----
For reference, here's my Motu 828 MkII -10dBu 1KHz sine loopback spectra
Attachments
Checking the unity gain noise and freq response with no GR but with the iso-transformer (Art isobox 10K:10K)
noval channel -3dB 55Hz and noise -69dBu
octal channel -3dB 85Hz and noise -82dBu
So I've picked up a ground loop on the noval with all the side chain hookup.
Interestingly the iso 10K:10K shows only a 5Hz worsening of the low freq extension. No sign of any issues with the dc offset present on it's outputs.
OK - some tweaks to do to improve the noise on the noval channel, and more detailed checks on the side chain, but not bad!
----
For reference, here's my Motu 828 MkII noise loopback spectra
noval channel -3dB 55Hz and noise -69dBu
octal channel -3dB 85Hz and noise -82dBu
So I've picked up a ground loop on the noval with all the side chain hookup.
Interestingly the iso 10K:10K shows only a 5Hz worsening of the low freq extension. No sign of any issues with the dc offset present on it's outputs.
OK - some tweaks to do to improve the noise on the noval channel, and more detailed checks on the side chain, but not bad!
----
For reference, here's my Motu 828 MkII noise loopback spectra
Attachments
Cripes!
Another 90degree twist on the utc a15 interstage on the offending noval channel and bingo!
50Hz -102dBu, 100Hz, -105dBu, 150Hz -92dBu
Unity gain Noisefloor drops to noval -82.5dBu - so now basically matches the octal -83.1dBu
(loopback is -84.3dBu)
That means each of the interstage appears to be oriented differently - I can see the label is different on each with respect to the pinout orientation, but I expected the 'pinout orientation' should be the same for both - but who knows 'what goes on'.
Whatever!
Anyway I'm thrilled - unity noise floors with all circuitry is REALLY GOOD on both channels.
That's base station #1 - complete.
It is stable and haven't find any particular wiring sensitivities.
Tap test is a little worse now but still well within the range of what I'd expect.
---
SO - only remaining problem is now the octal low freq extension some 15Hz higher for the octal.
Another 90degree twist on the utc a15 interstage on the offending noval channel and bingo!
50Hz -102dBu, 100Hz, -105dBu, 150Hz -92dBu
Unity gain Noisefloor drops to noval -82.5dBu - so now basically matches the octal -83.1dBu
(loopback is -84.3dBu)
That means each of the interstage appears to be oriented differently - I can see the label is different on each with respect to the pinout orientation, but I expected the 'pinout orientation' should be the same for both - but who knows 'what goes on'.
Whatever!
Anyway I'm thrilled - unity noise floors with all circuitry is REALLY GOOD on both channels.
That's base station #1 - complete.
It is stable and haven't find any particular wiring sensitivities.
Tap test is a little worse now but still well within the range of what I'd expect.
---
SO - only remaining problem is now the octal low freq extension some 15Hz higher for the octal.
Attachments
And here's the octal - bit better due to the lesser low freq extension in the area of schmutz.
50Hz -109dBu, 100Hz -94dBu, 150Hz -100dBu, 250Hz -102dBu
---
Harmonics at -10dBu input are basically the same across both channels, octal a touch higher 2nd harmonic.
2nd harmonic 50dB below fundamental, 3rd 47dB, 4th 67dB, 5th 66dB, higher vanishing low
I've yet to do a detail balancing of GR tube plate, cathode and finals cathode.
---
Looking at noise at no GR and with applied gain, they both add about as much broadband noise as gain applied - mostly 100Hz noise issue.
octal +15dB gain, noisefloor -70.3dBu (worsens by 12.6dB)
noval +15dB gain, noisefloor -70.5dBu (worsens by 12.2dB)
50Hz -109dBu, 100Hz -94dBu, 150Hz -100dBu, 250Hz -102dBu
---
Harmonics at -10dBu input are basically the same across both channels, octal a touch higher 2nd harmonic.
2nd harmonic 50dB below fundamental, 3rd 47dB, 4th 67dB, 5th 66dB, higher vanishing low
I've yet to do a detail balancing of GR tube plate, cathode and finals cathode.
---
Looking at noise at no GR and with applied gain, they both add about as much broadband noise as gain applied - mostly 100Hz noise issue.
octal +15dB gain, noisefloor -70.3dBu (worsens by 12.6dB)
noval +15dB gain, noisefloor -70.5dBu (worsens by 12.2dB)
Attachments
Now for some more GR testing ...
With -10dBu 1KHz sine input thru iso and
-> input attenuator at 'min attenuation'
-> output atten set for unity gain
-> release at slowest, attack at fastest
I set the Threshold for max GR : octal -10dB GR, noval -8dB GR
octal frequency response 'improves' to quite flat, as does noval - theres a little wobbling at the low end extension but looks very good overall.
** Now, freq response sweep test when unit is under GR is not really reliable imho
** But, as a general indicator of gross loss of freqs under GR, useful
Here's an example - this noval channel had a 'no GR' low -3dB 55Hz.
With -10dBu 1KHz sine input thru iso and
-> input attenuator at 'min attenuation'
-> output atten set for unity gain
-> release at slowest, attack at fastest
I set the Threshold for max GR : octal -10dB GR, noval -8dB GR
octal frequency response 'improves' to quite flat, as does noval - theres a little wobbling at the low end extension but looks very good overall.
** Now, freq response sweep test when unit is under GR is not really reliable imho
** But, as a general indicator of gross loss of freqs under GR, useful
Here's an example - this noval channel had a 'no GR' low -3dB 55Hz.
Attachments
Now to more extreme GR tests!
Setting input signal to +4dBu, 1Khz and Threshhold for 15dB GR, Level for 15dB applied gain
(then removing signal to show 'silence')
- CV achieved -19.2V
- noval -71.2 dBu noise floor, - octal -71.3 dBu noise floor
----> updated for balancing/debalancing, removing iso
So that's a deal worse than expected ie. mostly in the 100Hz, 150Hz noise components.
More work to be done there - however, this is around a worse case and is still well above my 'performance' cut off of -64dBu at silence with 15dB applied gain.
Summary So Far
-------------------
The Good
- unity gain noisefloor both channels
- THD low at moderate applied gain, and moderate at high applied gain
- GR easily achievable >20 dB both channels
- extended frequency response under GR both channels
- controls 'Threshhold', 'Attack', 'Release' don't add noise/instability at any setting
- stability with high GR, high applied gain
- virtually no PSU interactions between the channels
- virtually no audio cross talk between the channels
- 40hrs+ burn-in time so far and no smoke production or fuse melting!
The Bad
- noisefloor with applied gain needs further optimisation both channels by around 5dB
- 'no GR' low freq extension the octal channel needs further optimisation by around 10Hz
The Ugly
- build appears to be very critical of choice of input transformer
- some difference between octal and noval at high applied gain at i) noise freqs, ii) 2nd, 3rd harms
Summary
-----------
- so far it's all pretty good, just need to optimise noise with applied gain some
Setting input signal to +4dBu, 1Khz and Threshhold for 15dB GR, Level for 15dB applied gain
(then removing signal to show 'silence')
- CV achieved -19.2V
- noval -71.2 dBu noise floor, - octal -71.3 dBu noise floor
----> updated for balancing/debalancing, removing iso
So that's a deal worse than expected ie. mostly in the 100Hz, 150Hz noise components.
More work to be done there - however, this is around a worse case and is still well above my 'performance' cut off of -64dBu at silence with 15dB applied gain.
Summary So Far
-------------------
The Good
- unity gain noisefloor both channels
- THD low at moderate applied gain, and moderate at high applied gain
- GR easily achievable >20 dB both channels
- extended frequency response under GR both channels
- controls 'Threshhold', 'Attack', 'Release' don't add noise/instability at any setting
- stability with high GR, high applied gain
- virtually no PSU interactions between the channels
- virtually no audio cross talk between the channels
- 40hrs+ burn-in time so far and no smoke production or fuse melting!
The Bad
- noisefloor with applied gain needs further optimisation both channels by around 5dB
- 'no GR' low freq extension the octal channel needs further optimisation by around 10Hz
The Ugly
- build appears to be very critical of choice of input transformer
- some difference between octal and noval at high applied gain at i) noise freqs, ii) 2nd, 3rd harms
Summary
-----------
- so far it's all pretty good, just need to optimise noise with applied gain some
Next up - some time constant checks, guitar test and then ... music test, looking for 'thump'
----
- attack and release working as I expect
----
- guitar test is very fine
- in typical usage so far I'm seeing around 7dB GR and and have 12dB makeup for a 'silence' noise floor at around -73dBu which is good.
Not stellar like the Drip 175b build, with it's -81dBu at same gain, but respectable.
It does seem very smooth on the release - not hearing any lumpiness but I haven't pushed up the GR yet
I'm seeing a CV of around -10.6Vdc and the grid-cath difference at -13.8V - that's still about 2.3mA per phase I estimate (down from idle of 7.2mA or so) - plenty of room to go down yet!
My Motu guitar preamp/line out is pretty limited and can't really push out more than -15dBu or so.
So until I put on a proper preamp, I won't really get the serious GR - this build needs 0dBu or so solid because of the lack of input transformer step up.
But very nice indeed at 7dB GR - smooth, quiet on both, plenty of output - level is at around 10 o'clock so far for 12dB gain applied, -73dBu noise floor
No thump of any kind on this test - completely quiet and smooth.
I can hear the octal channel with a little less bottom end than the noval, as I would expect here.
But not a lot in it for guitar - certainly a little bottom boost would sort that out nicely. I just dialled out some top end and boosted a little - all good.
Apart from that, the octal and noval are pretty similar - at this stage, I seem to prefer the ....
OCTALS ;D 8);D 8) ;D 8)
Subjectively, the Drip 175b was more impressive at this stage - but not really a fair comparison as I can't dial up the full capability of the rca style build yet.
But the Drip noise performance with gain is just superb, so I can go to much higher levels of GR/makeup without getting too ... hummy
I would characterise the rca-style build as ... smooth, clear, quite transparent and with only a touch of velvet crunch in the line amp.
I would also say the Drip 175b is a classier sound - less crunch in the line amp, more 'glissando' quality in the compression.
The rca build, however seems to have a 'less lumpy' release.
I WOULD like to hear an rca style build with the Drip 175b Sowter transformers.
The pentode mode GR with stronger sidechain seems a very nice combo indeed.
ONE other thing - the large frame Edcor sound pretty good!
MUCH better than the xsm. AND they work 100% every time so far
Got all that ;D
-----
So far so good - this one is going much better than I expected in such a short time since I restarted 8) 8)
Another session or two trying to improve noise, a quick rebias for the changed noval 6BD6, AT7 tubes and finalise the heater resistor (for 5.9A ??? ???) yet to do.
As well as complete the utilities, of course - which I always do last.
----
- attack and release working as I expect
----
- guitar test is very fine
- in typical usage so far I'm seeing around 7dB GR and and have 12dB makeup for a 'silence' noise floor at around -73dBu which is good.
Not stellar like the Drip 175b build, with it's -81dBu at same gain, but respectable.
It does seem very smooth on the release - not hearing any lumpiness but I haven't pushed up the GR yet
I'm seeing a CV of around -10.6Vdc and the grid-cath difference at -13.8V - that's still about 2.3mA per phase I estimate (down from idle of 7.2mA or so) - plenty of room to go down yet!
My Motu guitar preamp/line out is pretty limited and can't really push out more than -15dBu or so.
So until I put on a proper preamp, I won't really get the serious GR - this build needs 0dBu or so solid because of the lack of input transformer step up.
But very nice indeed at 7dB GR - smooth, quiet on both, plenty of output - level is at around 10 o'clock so far for 12dB gain applied, -73dBu noise floor
No thump of any kind on this test - completely quiet and smooth.
I can hear the octal channel with a little less bottom end than the noval, as I would expect here.
But not a lot in it for guitar - certainly a little bottom boost would sort that out nicely. I just dialled out some top end and boosted a little - all good.
Apart from that, the octal and noval are pretty similar - at this stage, I seem to prefer the ....
OCTALS ;D 8);D 8) ;D 8)
Subjectively, the Drip 175b was more impressive at this stage - but not really a fair comparison as I can't dial up the full capability of the rca style build yet.
But the Drip noise performance with gain is just superb, so I can go to much higher levels of GR/makeup without getting too ... hummy
I would characterise the rca-style build as ... smooth, clear, quite transparent and with only a touch of velvet crunch in the line amp.
I would also say the Drip 175b is a classier sound - less crunch in the line amp, more 'glissando' quality in the compression.
The rca build, however seems to have a 'less lumpy' release.
I WOULD like to hear an rca style build with the Drip 175b Sowter transformers.
The pentode mode GR with stronger sidechain seems a very nice combo indeed.
ONE other thing - the large frame Edcor sound pretty good!
MUCH better than the xsm. AND they work 100% every time so far
Got all that ;D
-----
So far so good - this one is going much better than I expected in such a short time since I restarted 8) 8)
Another session or two trying to improve noise, a quick rebias for the changed noval 6BD6, AT7 tubes and finalise the heater resistor (for 5.9A ??? ???) yet to do.
As well as complete the utilities, of course - which I always do last.
