Redd 47 Pre with Drive+EQ DIY
- Thread starter alexc
- Start date
Help Support GroupDIY Audio Forum:
The there's the 'Orange86' preamp module tests.
First up is the 'unity gain noisefloor'
ie. setup RTAS to send -10dBu 1KHz sine to the unit, adjust feedback-gain rotary switch for -10dBu at the output transformer secondary, as visible on the RTA.
Then remove signal and look at the 'unity gain' spectrum -> noisefloor
Headline figure is -83.96dBu which is a mere 0.4dB worsening over loopback
Very, very good.
First up is the 'unity gain noisefloor'
ie. setup RTAS to send -10dBu 1KHz sine to the unit, adjust feedback-gain rotary switch for -10dBu at the output transformer secondary, as visible on the RTA.
Then remove signal and look at the 'unity gain' spectrum -> noisefloor
Headline figure is -83.96dBu which is a mere 0.4dB worsening over loopback
Very, very good.
Attachments
Next up is the -10dBu signal spectrum
Same as before but examine with signal present, looking at hum frequencies, hiss frequencies and harmonics of fundamental 1KHz tone applied
Shows quite low THD, as far as tube gear goes - mostly low order as expected with unity gain.
ie. RTAS -> Edcor WSX 10K:10K input traffo -> Orange86 -> Edcor XSM 15K:600 output traffo
Same as before but examine with signal present, looking at hum frequencies, hiss frequencies and harmonics of fundamental 1KHz tone applied
Shows quite low THD, as far as tube gear goes - mostly low order as expected with unity gain.
ie. RTAS -> Edcor WSX 10K:10K input traffo -> Orange86 -> Edcor XSM 15K:600 output traffo
Attachments
Next is same thing but some applied gain, in this case +13.4dB
THD increased, as expected with the reduction in feedback around the preamp due to setting the rotary-switch-feedback attenuator for more gain.
Still pretty good - you can also see more 'density' in the harmonic makeup as higher orders become more significant
THD increased, as expected with the reduction in feedback around the preamp due to setting the rotary-switch-feedback attenuator for more gain.
Still pretty good - you can also see more 'density' in the harmonic makeup as higher orders become more significant
Attachments
Then the same thing with signal removed, to show the +13dB noisefloor'
Hum frequences have increased some, as have the hiss frequencies. Highs not so much.
Headline figure is -75.8dBu which is a worsening of 8.5dB for +13.4dB gain.
That's pretty good - very good at this early stage, before any optimisation.
That's at the top of the second tier of my builds - after the Drip 175b, PMsix61 and a few others.
Hum frequences have increased some, as have the hiss frequencies. Highs not so much.
Headline figure is -75.8dBu which is a worsening of 8.5dB for +13.4dB gain.
That's pretty good - very good at this early stage, before any optimisation.
That's at the top of the second tier of my builds - after the Drip 175b, PMsix61 and a few others.
Attachments
Lastly, some freq response sweeps.
This one is -20dBu signal, +10dB gain applied.
You can see its very flat.
This measurement the 'mic' channel with it's utc O-8 + shield set for 250:15K impedance, x7.75 voltage or +17.8dB step up ratio.
With the line channel, even flatter.
Freq response is somewhat dependent on input level and applied gain - typically as these are increased, the -3dB points shift in, either side - making an 'upside down smiley face'.
Also, at higher levels, the hum noise starts showing up in the sweep as 'bumps' in the low end.
This one is -20dBu signal, +10dB gain applied.
You can see its very flat.
This measurement the 'mic' channel with it's utc O-8 + shield set for 250:15K impedance, x7.75 voltage or +17.8dB step up ratio.
With the line channel, even flatter.
Freq response is somewhat dependent on input level and applied gain - typically as these are increased, the -3dB points shift in, either side - making an 'upside down smiley face'.
Also, at higher levels, the hum noise starts showing up in the sweep as 'bumps' in the low end.
Attachments
Now all that shows the Orange86 is a good performer.
I have a russian 6J32P which is said to be 'equivalent' to an EF86 - it surely is a direct sub electrically but has some different performance in terms of the curves etc.
Also using a russian 6N5P as the driver. I talked about the bias I chose earlier in this thread.
I'll just spend a little more time on optimising noise performance and move on.
------------
Next up - testing the line amp part of the 'eq+drive' module I'm throwing together for this build.
Cheers
I have a russian 6J32P which is said to be 'equivalent' to an EF86 - it surely is a direct sub electrically but has some different performance in terms of the curves etc.
Also using a russian 6N5P as the driver. I talked about the bias I chose earlier in this thread.
I'll just spend a little more time on optimising noise performance and move on.
------------
Next up - testing the line amp part of the 'eq+drive' module I'm throwing together for this build.
Cheers
I have the line amp part of the eq+drive wired up and with basic testing done.
It is a push-pull GE nos 12B4A finals, with old stock RCA 12AU7 as a 'split load inverter' driving an Edcor XSM 15K:600. After that is a CAPI T-pad to the XLRs.
Feeding that is a grounded cathode gain stage with the other part of the AU7.
That stage is currently fed right off the little altronics 600:600 traffo which links the eq_drive module to the output traffo of the Orange86.
----
It all biases up pretty much exactly right. Haven't measured the gain but looks like at least +20dB, probably more, which is in the ballpark for the EQ makeup.
I can add more gain by biasing the push-pull for more current. I have 15mA+15mA on the finals at present, but I can go to 25mA+25mA on this tube with no trouble if necessary.
----
With the gain turned down a few clicks at the Orange86 feedback-atten rotary switch, and with the T-pad at the output turned down some, I can get unity gain.
The unity gain noise floor is very good - better than -80dBu right now, BUT there is some low freq rolloff at present, so that is making it look better than it is.
So, I need to track down where the low freq extension is being lost.
The little altronics traffo - it measures up good and flat when checked in this circuit with the line amp following, disconnected. Now to check more.
---
Confirmed no problem with the input transformer - looks like my 'step network', coupling the gain stage to the split load PI needs adjusticating - the 20nF is too low. And the resistors are probably too high, although the ratio is about right.
So - next up is to rework the step network.
When that's going and I have my response back, I'll keep moving forward to the front of them amp.
The real fun part is there - putting in the eq and drive controls.
Cheers
It is a push-pull GE nos 12B4A finals, with old stock RCA 12AU7 as a 'split load inverter' driving an Edcor XSM 15K:600. After that is a CAPI T-pad to the XLRs.
Feeding that is a grounded cathode gain stage with the other part of the AU7.
That stage is currently fed right off the little altronics 600:600 traffo which links the eq_drive module to the output traffo of the Orange86.
----
It all biases up pretty much exactly right. Haven't measured the gain but looks like at least +20dB, probably more, which is in the ballpark for the EQ makeup.
I can add more gain by biasing the push-pull for more current. I have 15mA+15mA on the finals at present, but I can go to 25mA+25mA on this tube with no trouble if necessary.
----
With the gain turned down a few clicks at the Orange86 feedback-atten rotary switch, and with the T-pad at the output turned down some, I can get unity gain.
The unity gain noise floor is very good - better than -80dBu right now, BUT there is some low freq rolloff at present, so that is making it look better than it is.
So, I need to track down where the low freq extension is being lost.
The little altronics traffo - it measures up good and flat when checked in this circuit with the line amp following, disconnected. Now to check more.
---
Confirmed no problem with the input transformer - looks like my 'step network', coupling the gain stage to the split load PI needs adjusticating - the 20nF is too low. And the resistors are probably too high, although the ratio is about right.
So - next up is to rework the step network.
When that's going and I have my response back, I'll keep moving forward to the front of them amp.
The real fun part is there - putting in the eq and drive controls.
Cheers
Doing some clacs on the step network shows 0.1uF is good for a -3dB point of around 30Hz or so.
Making the change confirms that the freq response there is restored.
So the step network is a 'frequency conscious voltage divider' with a cutoff at around 30Hz.
This reduces my dc at the grid of the split load phase inverter - now down to around +61Vdc
I'm doing that so as to maintain a low enough cathode bias so as not to exceed the au7 max heater-cathode voltage of 100V for the dc component, 200V for the dc+signal total.
I can't just raise the heater reference because the heater is dc, ground referenced.
The plate at the preceeding stage is at 173V so the step network drops it to 61V dc without compromising the low-freq extension too much.
That still gives me plenty of signal swing at the grid of the PI - around 100Vpp without approaching the grid limit. PI gain was around x0.87
I need only around 20Vpp to fully drive the finals, with their cathodes now at 11V or so.
So - plenty of 'crunch' potential, should one desire. (which I don't)
Also lots of scope for feedback to smooth the schmutz out.
Making the change confirms that the freq response there is restored.
So the step network is a 'frequency conscious voltage divider' with a cutoff at around 30Hz.
This reduces my dc at the grid of the split load phase inverter - now down to around +61Vdc
I'm doing that so as to maintain a low enough cathode bias so as not to exceed the au7 max heater-cathode voltage of 100V for the dc component, 200V for the dc+signal total.
I can't just raise the heater reference because the heater is dc, ground referenced.
The plate at the preceeding stage is at 173V so the step network drops it to 61V dc without compromising the low-freq extension too much.
That still gives me plenty of signal swing at the grid of the PI - around 100Vpp without approaching the grid limit. PI gain was around x0.87
I need only around 20Vpp to fully drive the finals, with their cathodes now at 11V or so.
So - plenty of 'crunch' potential, should one desire. (which I don't)
Also lots of scope for feedback to smooth the schmutz out.
Thanks - this build is already sounding pretty good.
My 'philology' for colored tube stuff is tube stages biased quite clean, but plenty of them!
In this case, there's quite a few of atten/gain stages throughout, so the 'crunch' can layer on gradually if you want it. I'm just listening and getting a feel for the textures.
Right now I can get quite a bit of crunch already
- I'm using the Motu guitar preamp and sending to the build line input with it's edcor wsm.
****
I'm liking - max pad, min input atten (least feedback), output atten for typical 0dBu
Has a little crunch and quite a lot of compression (?) - that's what it sounds like to me.
****
That is all without the eq or drive section. I'm not even using the pentode yet
I actually changed to 6S4A finals - they have more gain at less current and I was seeing a better freq response in this app. They are low plate Z but no as low as the 12B4A.
I'm going to do some feedback from the finals back to the PI input, similar to fender amps. That should take out some gain and smooth out things a little.
-----
Don't know what the Motu 828MkII dBFS to dBu relationship is, really.
It has max output/input of around 11dBu, 7.77Vpp beyond which the distortion rapidly goes bonkers for the last 1dB or so to max levels.
I guess that means 0 dBFS (max digital levels) whould be 12dBu.
I'll do some further checks and post it what I think.
Cheers
My 'philology' for colored tube stuff is tube stages biased quite clean, but plenty of them!
In this case, there's quite a few of atten/gain stages throughout, so the 'crunch' can layer on gradually if you want it. I'm just listening and getting a feel for the textures.
Right now I can get quite a bit of crunch already
- I'm using the Motu guitar preamp and sending to the build line input with it's edcor wsm.****
I'm liking - max pad, min input atten (least feedback), output atten for typical 0dBu
Has a little crunch and quite a lot of compression (?) - that's what it sounds like to me.
****
That is all without the eq or drive section. I'm not even using the pentode yet
I actually changed to 6S4A finals - they have more gain at less current and I was seeing a better freq response in this app. They are low plate Z but no as low as the 12B4A.
I'm going to do some feedback from the finals back to the PI input, similar to fender amps. That should take out some gain and smooth out things a little.
-----
Don't know what the Motu 828MkII dBFS to dBu relationship is, really.
It has max output/input of around 11dBu, 7.77Vpp beyond which the distortion rapidly goes bonkers for the last 1dB or so to max levels.
I guess that means 0 dBFS (max digital levels) whould be 12dBu.
I'll do some further checks and post it what I think.
Cheers
Attachments
Yep - this is one super kick ass guitar pre, alight 8)
Not too ice-pick 'twin reverb'-y, not too bassy - seems to have a 'goldilocks' mid range thick crunch.
'Just the way I like it'.
Right now, actually doesn't *need* anything else.
But I've never let necessity get in the way of a good, old fashioned feature overkill.
----
Seeing a basic noise floor at these levels, which is actually +10dB gain applied, of around -77.5dBu which is pretty good.
ie 6dBu 'rms noise 20-20KHz added for 10dB of added gain.
Not great, but very good - especially considering signal is passing thru 4 transformers, 5 tubes and 3 attenuators AND I haven't yet begun to tweak.
The hiss, mid range 500Hz - 2KHz, are quite low.
Here's a spectrum with signal at these typical settings.
Not too ice-pick 'twin reverb'-y, not too bassy - seems to have a 'goldilocks' mid range thick crunch.
'Just the way I like it'.
Right now, actually doesn't *need* anything else.
But I've never let necessity get in the way of a good, old fashioned feature overkill.
----
Seeing a basic noise floor at these levels, which is actually +10dB gain applied, of around -77.5dBu which is pretty good.
ie 6dBu 'rms noise 20-20KHz added for 10dB of added gain.
Not great, but very good - especially considering signal is passing thru 4 transformers, 5 tubes and 3 attenuators
AND I haven't yet begun to tweak.The hiss, mid range 500Hz - 2KHz, are quite low.
Here's a spectrum with signal at these typical settings.
Attachments
So I did the 'toroid twist' and lid on test.
My +10dB gain applied noise floor just improved to -81.0dBu 8) 8)
So that's 3.4dB noise added for 10dB of gain - I think I can squeeze another dB yet.
The unity gain noise added is almost nil.
That's about as good as the Drip175b build, which is #1 for quiet.
AWESOME!!!
My +10dB gain applied noise floor just improved to -81.0dBu 8) 8)
So that's 3.4dB noise added for 10dB of gain - I think I can squeeze another dB yet.
The unity gain noise added is almost nil.
That's about as good as the Drip175b build, which is #1 for quiet.
AWESOME!!!
So that's the line amp done nicely. Time to move on.
First, I'm adding a pot before the au7 - this will be something like a 12dB attenuator, and will be called 'Drive'.
In front of that will be the pentode. For now, given I have plenty of gain in hand, I'm going to work up the eq circuit first.
Eq can be a little tricky - at least for me! So I'll get that right before the pentode complication.
---
What I have in mind is a fender style tone stack with twists as well as an extra HPF, LPF filters.
The HPF will probably sit in front of the AU7 in the line amp.
I'm going to extend the existing Orange86 LPF from a fixed freq to a rotary switch of 6 positions.
This filter is on the preamp feedback-atten switch, so just a simple build out there.
---
The tone stack is going to have the expected hi, mid, low pots as well as an 'eq blend' pot which will mix the dry and Eq'd signal.
The only other twist is a 'shift' switch/pot which will vary the tone stack low and high freq points.
And, of course, the tone stack will be defeatable, substituting a level compensating resistor.
-----
So - that's quite a challenge I think.
Best get cracking!
First, I'm adding a pot before the au7 - this will be something like a 12dB attenuator, and will be called 'Drive'.
In front of that will be the pentode. For now, given I have plenty of gain in hand, I'm going to work up the eq circuit first.
Eq can be a little tricky - at least for me! So I'll get that right before the pentode complication.
---
What I have in mind is a fender style tone stack with twists as well as an extra HPF, LPF filters.
The HPF will probably sit in front of the AU7 in the line amp.
I'm going to extend the existing Orange86 LPF from a fixed freq to a rotary switch of 6 positions.
This filter is on the preamp feedback-atten switch, so just a simple build out there.
---
The tone stack is going to have the expected hi, mid, low pots as well as an 'eq blend' pot which will mix the dry and Eq'd signal.
The only other twist is a 'shift' switch/pot which will vary the tone stack low and high freq points.
And, of course, the tone stack will be defeatable, substituting a level compensating resistor.
-----
So - that's quite a challenge I think.
Best get cracking!
simonsez
Well-known member
Great built Alex! Where did you get those VU meter?
Thanks!
VU meter is a relatively cheap one from Altronics in Australia. I'm using it as I have a couple left over from years ago. This is a 'use up spare parts' build
Not internally rectified. It's OK but is a typical lightweight movement - just drifts around.
However, with the JLM meter buffer, it works OK.
The nissei costs about the same on ebay and is much better movement and much better construction.
VU meter is a relatively cheap one from Altronics in Australia. I'm using it as I have a couple left over from years ago. This is a 'use up spare parts' build
Not internally rectified. It's OK but is a typical lightweight movement - just drifts around.
However, with the JLM meter buffer, it works OK.
The nissei costs about the same on ebay and is much better movement and much better construction.
I took a standard 'Marshall' tone stack, which is the same as the 'Fender' tone stack but with the mid range cap taken from the mid pot wiper, not mid pot entry.
I used 470pF, 100nF and 22nF caps. Did the sims with the 'Duncan Tone Stack Simulator' software application (freeby). Tested/tweaked out of circuit till happy, then added to my build.
At the bottom of the stack, before the mid-pot end ground connection, I put a switch and a 1M log pot. This is the 'blend' control - when resistance is added before the ground, the eq effect is reduced.
With 1M pot, you can take out basically all eq effect, although the attenuation is still there.
An extra switch to open circuit that point is the same as setting the pot to 1M.
At the 'slope' resistor, I added a 500K linear pot. This is the 'shift' control. Adding resistance shifts the operating points of the eq to the low end. This gives me around 300Hz shift.
-----
At some typical setting, looks like around -12dB attenuation.
----
Now to do some listening tests.
I used 470pF, 100nF and 22nF caps. Did the sims with the 'Duncan Tone Stack Simulator' software application (freeby). Tested/tweaked out of circuit till happy, then added to my build.
At the bottom of the stack, before the mid-pot end ground connection, I put a switch and a 1M log pot. This is the 'blend' control - when resistance is added before the ground, the eq effect is reduced.
With 1M pot, you can take out basically all eq effect, although the attenuation is still there.
An extra switch to open circuit that point is the same as setting the pot to 1M.
At the 'slope' resistor, I added a 500K linear pot. This is the 'shift' control. Adding resistance shifts the operating points of the eq to the low end. This gives me around 300Hz shift.
-----
At some typical setting, looks like around -12dB attenuation.
----
Now to do some listening tests.
Well that sounds as I expect - I'm still getting around -80dBu noisefloor at typical settings, which is around +17dB gain now - that is around 12dB of eq makeup and 5db of general gain.
No problem at all - output atten is still at around 1/2 way. The pad is engaged and the input atten fairly low at 1/3 of the travel so as not to push the front too hard.
Not getting any audible hum, as I would expect at those measured noise levels.
There is more hiss now - quite a bit going on in those mid freqs, so it is to be expected.
Also, even though the eq is driven from a low Z, the tone stack is quite high Z, with 500K and 1M pots in there. It would probably pay to scale down the high value pots and so on - likely not on the cards for me, though.
-----
Next up - the 'drive' section.
A 6BJ6, my favourite little pentode - it has a lot of push for such a little guy using next to no heater.
Can pass a fair current too - I've used it at 15mA total plate/screen no probs.
The input to this is the tone stack, which ends in a 'Pre' pot of 1M with attenuation all the way to nil.
So that's a fairly hiZ driving source to it's grid.
I will bias it up fairly cold, so as to take a goodly voltage swing, maybe 10Vpp or so before grid limiting. Gain will be low-ish and the output will likely go to a divider then to the 'Drive' pot at the front of the line amp.
Twists will be a 'mode' pot for transitioning smoothly from triode to pentode mode as well as anything else I can do - thinking some messing with the screen in 'pentode' mode.
The idea is for maybe 24dB of gain, then to divide down to 10dB or so before hitting the line amp.
I'll see how it goes.
So far, so good
No problem at all - output atten is still at around 1/2 way. The pad is engaged and the input atten fairly low at 1/3 of the travel so as not to push the front too hard.
Not getting any audible hum, as I would expect at those measured noise levels.
There is more hiss now - quite a bit going on in those mid freqs, so it is to be expected.
Also, even though the eq is driven from a low Z, the tone stack is quite high Z, with 500K and 1M pots in there. It would probably pay to scale down the high value pots and so on - likely not on the cards for me, though.
-----
Next up - the 'drive' section.
A 6BJ6, my favourite little pentode - it has a lot of push for such a little guy using next to no heater.
Can pass a fair current too - I've used it at 15mA total plate/screen no probs.
The input to this is the tone stack, which ends in a 'Pre' pot of 1M with attenuation all the way to nil.
So that's a fairly hiZ driving source to it's grid.
I will bias it up fairly cold, so as to take a goodly voltage swing, maybe 10Vpp or so before grid limiting. Gain will be low-ish and the output will likely go to a divider then to the 'Drive' pot at the front of the line amp.
Twists will be a 'mode' pot for transitioning smoothly from triode to pentode mode as well as anything else I can do - thinking some messing with the screen in 'pentode' mode.
The idea is for maybe 24dB of gain, then to divide down to 10dB or so before hitting the line amp.
I'll see how it goes.
So far, so good
Attachments
I have the pentode all wired up now as well as the 'mode' pot - just deciding on the atten at the output. I have 20-24dB there now and will like reduce it by 6dB to 18dB on a pot.
It does come out pretty crunchy - plenty of attenuation steps to find a desired level of THD.
I can dial in quite low THD all the way to a LOT.
At unity gain, with typical eq setting, I'm seeing around 69dBu noise floor with the pentode in circuit.
The 50Hz component has jumped a fair bit, as one would expect and the mid-range too.
Without, it remains at 83dBu, very close to loopback.
----
I do think I'll put a bypass switch on the whole pentode section to provide a 'clean' setting.
That line amp as it is has plenty of gain and sounds quite nice alone.
Am doing up the HPF now too - simple rotary switch varying the value of the coupling cap after the pentode stage - 6 steps will do I think.
So - some last couple of things to do there and I start bashing the front panel.
I may also replace the wsm 10K:10K input with my last utc a-series - I'll see which one works best there once I button all the other stuff up.
Here's a pic of how the front panel is shaping up - pretty busy, no? ;D
-- updated panel
- removed post EQ 'Saturation', added line amp 'Feedback' control
- changed filters to a single 'HiCut'
It does come out pretty crunchy - plenty of attenuation steps to find a desired level of THD.
I can dial in quite low THD all the way to a LOT.
At unity gain, with typical eq setting, I'm seeing around 69dBu noise floor with the pentode in circuit.
The 50Hz component has jumped a fair bit, as one would expect and the mid-range too.
Without, it remains at 83dBu, very close to loopback.
----
I do think I'll put a bypass switch on the whole pentode section to provide a 'clean' setting.
That line amp as it is has plenty of gain and sounds quite nice alone.
Am doing up the HPF now too - simple rotary switch varying the value of the coupling cap after the pentode stage - 6 steps will do I think.
So - some last couple of things to do there and I start bashing the front panel.
I may also replace the wsm 10K:10K input with my last utc a-series - I'll see which one works best there once I button all the other stuff up.
Here's a pic of how the front panel is shaping up - pretty busy, no? ;D
-- updated panel
- removed post EQ 'Saturation', added line amp 'Feedback' control
- changed filters to a single 'HiCut'
