Measurement Software - 'Room EQ Wizard'
- Thread starter alexc
- Start date
Help Support GroupDIY Audio Forum:
ruffrecords
Well-known member
So cool, pity it is mack only.
Cheers
ian
Cheers
ian
So I have re biased the el34pp amp stage 1 voltage amp - currently an EH 6922.
I've gone for a typical set up, I think : low-ish plate supply of +245V, high-ish plate load of 27K, moderate bias of +2.75V.
It's pretty happy in there - with a 5.2Vpp max input before h3+ starts accelerating upwards ....
Gain is as expected, +26.6dB and output is 111Vp.
Nice spectra - very big signal-hum ratio of > 110dB - probably my best yet - certainly since I began keeping detailed records!
So this is the current baseline for the 6922 mas a plain old grounded cathode gain stage.
I've gone for a typical set up, I think : low-ish plate supply of +245V, high-ish plate load of 27K, moderate bias of +2.75V.
It's pretty happy in there - with a 5.2Vpp max input before h3+ starts accelerating upwards ....
Gain is as expected, +26.6dB and output is 111Vp.
Nice spectra - very big signal-hum ratio of > 110dB - probably my best yet - certainly since I began keeping detailed records!
So this is the current baseline for the 6922 mas a plain old grounded cathode gain stage.
Attachments
It's interesting to note that I measured 'better' figures running at a higher plate voltage, but its probably not recommended to use 330V plate supply 8) for too long.
The robust EH6922 might be fine with it, but I plan to put some nos in here and they are probably somewhat more delicate!
The robust EH6922 might be fine with it, but I plan to put some nos in here and they are probably somewhat more delicate!
And to go along with that quite nice EH6922 gain stage, here is my rejiggered 6SN7 phase inverter setup.
Basically its 27K plate loads running at 6.75mA each side, with a+6.42V bias for plates at 295V off of a 470V supply.
I should knock the supply down a little to the rated 450V max
And I could really dial in to the best THD etc. But this is pretty good - the balance is set for 'neutral' - just ensured equal dc conditions, not 'dialled in for lowest THD' .. for now!
It's humming along nicely - with no global feedback connected, the el34s of the sed wingedC variety (circa 2010 or so) it needs to do 74Vpp output. Manages is nicely ... with 0.2% THD+N
I could do some more here - I have yet to do the tube rolling - this is my standard tung-sol (re-issue) 6sn7.
Not too bad!
Basically its 27K plate loads running at 6.75mA each side, with a+6.42V bias for plates at 295V off of a 470V supply.
I should knock the supply down a little to the rated 450V max
And I could really dial in to the best THD etc. But this is pretty good - the balance is set for 'neutral' - just ensured equal dc conditions, not 'dialled in for lowest THD' .. for now!It's humming along nicely - with no global feedback connected, the el34s of the sed wingedC variety (circa 2010 or so) it needs to do 74Vpp output. Manages is nicely ... with 0.2% THD+N
I could do some more here - I have yet to do the tube rolling - this is my standard tung-sol (re-issue) 6sn7.
Not too bad!
Attachments
And here is the speaker terminals at 30Wrms into 8ohm load box - no global negative feedback .. yet
Freq response is -0.8dB at 20Hz and -3dB at 24.4KHz
This monobloc is basically optimised and is my first 'baseline' performance scenario - the 'open loop' condition.
- less than 3% THD+N at full rated power - tick
- hum floor approaching -95 dBu - tick
- greater than 90dB signal-hum margin at the speaker terms at full rated power - tick
- mostly H2 at full rated power - tick
- less than 1dB droop at LF, HF for full rated power - tick
- phase shift measured - tick
I'll audition it here for a while, ruminate over my results and decide on whether to give it another round of VA + PI optimisation.
Happy days - it's a long, long way from where it began the 'powered by Room EQ Wizard (REQ)' tear down
It's like my old Merc SL - it hums and purrs after it's been to the mechanics. THEN you realise how poorly it was running beforehand
And yes - it sounds good enough and quiet enough to leave with no global negative feedback. However, I'll be applying something like 6dB to 12dB shortly. For hifi, I do like the low THD and increased damping factor that nfb provides
Freq response is -0.8dB at 20Hz and -3dB at 24.4KHz
This monobloc is basically optimised and is my first 'baseline' performance scenario - the 'open loop' condition.
- less than 3% THD+N at full rated power - tick
- hum floor approaching -95 dBu - tick
- greater than 90dB signal-hum margin at the speaker terms at full rated power - tick
- mostly H2 at full rated power - tick
- less than 1dB droop at LF, HF for full rated power - tick
- phase shift measured - tick
I'll audition it here for a while, ruminate over my results and decide on whether to give it another round of VA + PI optimisation.
Happy days - it's a long, long way from where it began the 'powered by Room EQ Wizard (REQ)' tear down
It's like my old Merc SL - it hums and purrs after it's been to the mechanics. THEN you realise how poorly it was running beforehand
And yes - it sounds good enough and quiet enough to leave with no global negative feedback. However, I'll be applying something like 6dB to 12dB shortly. For hifi, I do like the low THD and increased damping factor that nfb provides
Attachments
ruffrecords
Well-known member
Off to a good start on finalizing these el34pp monoblocs
Here's the latest spectra at the speaker terms of unit B with 15.5Vrms into 8ohm load box (30Wrms).
I've got about 7dB nfb connected - THD is very low now and noise floor is super low.
Best perf ever by a long shot. So, this is about where I'll stop.
Good news is that unit A and unit B are very, sery similar in performance - each is setup identically and the tubes are same types : eh 6922, tungsol (reissue) 6SN7, sed el34.
I'll do some more checks of the nfb network and what not then button them up and do the 'final' balance and bias on each.
THEN I'll do the benchmark snaps.
Further tube rolling at that point could be fun - it's where I wanted to be a couple of months ago!
Here's the latest spectra at the speaker terms of unit B with 15.5Vrms into 8ohm load box (30Wrms).
I've got about 7dB nfb connected - THD is very low now and noise floor is super low.
Best perf ever by a long shot.
So, this is about where I'll stop. Good news is that unit A and unit B are very, sery similar in performance - each is setup identically and the tubes are same types : eh 6922, tungsol (reissue) 6SN7, sed el34.
I'll do some more checks of the nfb network and what not then button them up and do the 'final' balance and bias on each.
THEN I'll do the benchmark snaps.
Further tube rolling at that point could be fun - it's where I wanted to be a couple of months ago!
Attachments
So, at full rated power of 30Wrms into 8ohms, I get :
1KHz sine input
0.187% THD+N -> 0.176% H2, 0.04% H3+
-99dBu hum floor
105dB signal-noise-ratio
and
freq resp +0.1dB at 20Hz (-1dB 10Hz) and -1.1dB 20KHz (-3dB at 46KHz)
phase resp +1deg at 20Hz and +6deg at 20KHz
So that's pretty good! ;D ;D
1KHz sine input
0.187% THD+N -> 0.176% H2, 0.04% H3+
-99dBu hum floor
105dB signal-noise-ratio
and
freq resp +0.1dB at 20Hz (-1dB 10Hz) and -1.1dB 20KHz (-3dB at 46KHz)
phase resp +1deg at 20Hz and +6deg at 20KHz
So that's pretty good! ;D ;D
Attachments
And here's what I went for in for by way of the buttoned up L'Amps 30W el34pp monoblocs :
- 6DJ8 amperex holland orange mid 60s nos
Little bit higher THD+N at 0.424%, and little bit better signal-hum margin.
I like the harmonics spectra a bit more
Listening is very nice indeed - a little less 'brash' than the eh 6922, which is about a dB or so louder. The amperex seems more 'touch sensitive' on the guitar.
So - super nice sound, inaudible hum, even on the big 15" speaker.
BACK into service they are, installed and happy once again as kings of my lounge room
- 6DJ8 amperex holland orange mid 60s nos
Little bit higher THD+N at 0.424%, and little bit better signal-hum margin.
I like the harmonics spectra a bit more
Listening is very nice indeed - a little less 'brash' than the eh 6922, which is about a dB or so louder. The amperex seems more 'touch sensitive' on the guitar.
So - super nice sound, inaudible hum, even on the big 15" speaker.
BACK into service they are, installed and happy once again as kings of my lounge room
Attachments
BACK again to the bigSEamp and my optimisation efforts ...
As they say, it all starts in the first tube position
Here is the spectra for my current baseline, tube position 1 - Mullard RI AX7, triode A
- front panel instrument jack (non isolated) with guitar 1.5Vpp input typical (strat wit stacked humbuckers)
Lower trace - the noise floor spectra - that's with the guitar volume pot fully down. ie. what you hear with everything connected and wide open, except the guitar
Upper tracer - the cumulative 'peak' or 'historical' trace in red, over 30s or so of playing, with +34.4dB gain
Typical signal-hum margin of the guitar input is pretty good at 89 dB or so. The pickups are Dimarzio virtual vintage, stacked humbuckers with low noise and 'vintage tone'
Its a very quiet, yet fairly hot pickup signal. Really wonderful pickups.
The AX7 is doing around +33.4dB of gain into a hiZ load - (running at 3.24% THD for a nominal 1Vpp input from sig gen at 1KHz)
This what I need to 'beat' : 89 dB of signal-hum margin, ideally at a lower intrinsic THD
The JJ5751s should be here in a couple of days
---
Meanwhile, I tested the effect of isolating the switchcraft jacks - actually is a bit noisier than the non-isolating scenario.
No changes needed - the jacks can remain connected to chassis ie. non-isolating. Also verified the other jacks aren't contributing to noise etc. All good, no improvements to be found there.
As they say, it all starts in the first tube position
Here is the spectra for my current baseline, tube position 1 - Mullard RI AX7, triode A
- front panel instrument jack (non isolated) with guitar 1.5Vpp input typical (strat wit stacked humbuckers)
Lower trace - the noise floor spectra - that's with the guitar volume pot fully down. ie. what you hear with everything connected and wide open, except the guitar
Upper tracer - the cumulative 'peak' or 'historical' trace in red, over 30s or so of playing, with +34.4dB gain
Typical signal-hum margin of the guitar input is pretty good at 89 dB or so. The pickups are Dimarzio virtual vintage, stacked humbuckers with low noise and 'vintage tone'
Its a very quiet, yet fairly hot pickup signal. Really wonderful pickups.
The AX7 is doing around +33.4dB of gain into a hiZ load - (running at 3.24% THD for a nominal 1Vpp input from sig gen at 1KHz)
This what I need to 'beat' : 89 dB of signal-hum margin, ideally at a lower intrinsic THD
The JJ5751s should be here in a couple of days
---
Meanwhile, I tested the effect of isolating the switchcraft jacks - actually is a bit noisier than the non-isolating scenario.
No changes needed - the jacks can remain connected to chassis ie. non-isolating. Also verified the other jacks aren't contributing to noise etc. All good, no improvements to be found there.
Attachments
And here is the same thing, but with input from sig gen, debalancing via some 15K/15K traffo, offset at the sig-gen by -6dB (normalising out the debalancing 6dB boost) for same resultant THD, and injected into the front panel 'instrument' jack.
Mullard reissue (2010) AX7 - quite consistent across a couple of tubes. This is my standard current production AX7 which I use in a number of my builds.
It's quite a balanced spectra and a nice performance. At this moment, I think it's going to be hard to beat.
So far, only the eh ay7 comes close. How close?
Mullard reissue (2010) AX7 - quite consistent across a couple of tubes. This is my standard current production AX7 which I use in a number of my builds.
It's quite a balanced spectra and a nice performance. At this moment, I think it's going to be hard to beat.
So far, only the eh ay7 comes close. How close?
Attachments
Taking advantage of the eh ay7 low, low thd here, you can increase the input slightly higher than the nominal 1Vpp I use for testing a high mu stage without blowing the thd figure ...
Here is the eh ay7 with around 1.8Vpp to the grid - now the signal-hum margin picks up to really good figures .. 94.3 dB and rising with any further 'strength' in the input signal.
Like a good stock pick!
Like I said, in this particular position, for a guitar/bass with a solid to hot output level, say 1.5Vpp to 2Vpp or even 2.5Vpp to 3Vpp at the outside .. is a good match for this tube and will turn in lowest in class thd.
All the while, keeping it real with a signal-hum margin approaching 100 dB on the louder stuff.
And a real-world 'silence' floor of approaching -100dBu
The common eh ay7 is a hard act to beat full stop.
The eh 6922 can do it, but not with a low-ish input signal. That tube needs some proper signal for it to really get in 'the zone'.
...
In the end, it's the 'acuity' of the match between the particular source and the triode's setup that make the difference from 'that's pretty good' to 'eye-popping moar please!' performance.
I use hot pups, I use the bass as well as guitar alot, so I favour the AY7 nearly every time - quiet and low distortion - your typical 'unhyp-ped' setup.
The AY7 also works well with hot pedals, as it can take quite a lot of input signal without getting nasty!
All that noted, ... for lower output pups, perhaps of varying noisiness, I think the AX7 takes the edge with it's ability to 'cut thru' a bit more, along with a still quite decent hum floor. That ol' thd goes up pretty quick though, so a tame signal is best unless you want to go 'H3+' in a more pronounced way.
Horses for courses of courses! (of courses!)
So - I await the JJ 5751 which I hope will mak shau
Here is the eh ay7 with around 1.8Vpp to the grid - now the signal-hum margin picks up to really good figures .. 94.3 dB and rising with any further 'strength' in the input signal.
Like a good stock pick!
Like I said, in this particular position, for a guitar/bass with a solid to hot output level, say 1.5Vpp to 2Vpp or even 2.5Vpp to 3Vpp at the outside .. is a good match for this tube and will turn in lowest in class thd.
All the while, keeping it real with a signal-hum margin approaching 100 dB on the louder stuff.
And a real-world 'silence' floor of approaching -100dBu
The common eh ay7 is a hard act to beat full stop.
The eh 6922 can do it, but not with a low-ish input signal. That tube needs some proper signal for it to really get in 'the zone'.
...
In the end, it's the 'acuity' of the match between the particular source and the triode's setup that make the difference from 'that's pretty good' to 'eye-popping moar please!' performance.
I use hot pups, I use the bass as well as guitar alot, so I favour the AY7 nearly every time - quiet and low distortion - your typical 'unhyp-ped' setup.
The AY7 also works well with hot pedals, as it can take quite a lot of input signal without getting nasty!
All that noted, ... for lower output pups, perhaps of varying noisiness, I think the AX7 takes the edge with it's ability to 'cut thru' a bit more, along with a still quite decent hum floor. That ol' thd goes up pretty quick though, so a tame signal is best unless you want to go 'H3+' in a more pronounced way.
Horses for courses of courses! (of courses!)
So - I await the JJ 5751 which I hope will mak shau
Attachments
I tried the JJ 5751 - quite good really
I bought 3 tubes, selected for low microphonics.
They all test very good for uniformity across the 3 tubes and seemed to have a consistent, and small difference in performance between triode 1 and 2 - I tested both triodes with an identical setup.
They come out better performing than my previous modern manufactured ax7, the Mullard 12AX7.
About 1 to 2 dB better on noise floor, in fact - which brings them into the similar perf as my fav EH AY7.
That's no mean feat!
The JJ 5751 harmonics are much nicer H2 than the Mullard RI ax7 - they have far less H3 and H3+.
So much so, in fact that the JJ 5751 closely resembles the EH ay7 spectra - low THD levels, mostly H2_
The gain is only 1 to 2 dB less than the ax7, and is well higher than the ay7, at around 3 to 4 dB more.
The other thing that stands out, is the transition from low THD - predominantly H2, to increasing H3, H3+.
It is *smoother* in this 5751 than the eh ay7.
The eh ay7 has low THD, all H2, until it borks - then a lot more of H3+, going to many orders.
The jj 5751 is transitioning more smoothly from H2 to H3 and H3+.
That's quite nice performance from the JJ 5751 - very quiet, very H2 and a smooth H2-H3 transitioning.
In use, it really works well as the position #1 in my signal chain, where I want as much signal-noise margin as possible, with low THD, H2 with a nominal 1.5Vpp or so input.
Nice and spangly top end, controlled mid range and bottom - good touch sensitivity and clarity. Digging in makes a nice growl too.
I'm finding myself hearing more from my guitar which surprises me - in a good way.
So - my new best friend and top of the charts in an ax7 role .. is the JJ 5751
I bought 3 tubes, selected for low microphonics.
They all test very good for uniformity across the 3 tubes and seemed to have a consistent, and small difference in performance between triode 1 and 2 - I tested both triodes with an identical setup.
They come out better performing than my previous modern manufactured ax7, the Mullard 12AX7.
About 1 to 2 dB better on noise floor, in fact - which brings them into the similar perf as my fav EH AY7.
That's no mean feat!
The JJ 5751 harmonics are much nicer H2 than the Mullard RI ax7 - they have far less H3 and H3+.
So much so, in fact that the JJ 5751 closely resembles the EH ay7 spectra - low THD levels, mostly H2
_The gain is only 1 to 2 dB less than the ax7, and is well higher than the ay7, at around 3 to 4 dB more.
The other thing that stands out, is the transition from low THD - predominantly H2, to increasing H3, H3+.
It is *smoother* in this 5751 than the eh ay7.
The eh ay7 has low THD, all H2, until it borks - then a lot more of H3+, going to many orders.
The jj 5751 is transitioning more smoothly from H2 to H3 and H3+.
That's quite nice performance from the JJ 5751 - very quiet, very H2 and a smooth H2-H3 transitioning.
In use, it really works well as the position #1 in my signal chain, where I want as much signal-noise margin as possible, with low THD, H2 with a nominal 1.5Vpp or so input.
Nice and spangly top end, controlled mid range and bottom - good touch sensitivity and clarity. Digging in makes a nice growl too.
I'm finding myself hearing more from my guitar which surprises me - in a good way.
So - my new best friend and top of the charts in an ax7 role .. is the JJ 5751
ruffrecords
Well-known member
A while back I did a lot of measurements of different brands of 6CG7 (the B9A version of the 6SN7). I discovered there was a direct relationship between the length (height) of the plate and the tube's distortion. The longer the plate the lower the distortion. I have done a smaller number of experiments with different brands of 12AX7 with the same results. It might be worth looking at the the length of the plates you have tested and seeing if this relationship holds true for them.
Cheers
Ian
Cheers
Ian
That's very interesting, Ian - here's the lengths I mersure :
JJ 5751 is a short plate at 9mm
EH AY7 is 12mm.
Mullard RI AX7is 17mm
The 5751 is very close in THD to the AY7 - both are low. The Mullard is higher in THD but still quite low compared to a bunch of other ax7 that I have.
JJ 5751 is a short plate at 9mm
EH AY7 is 12mm.
Mullard RI AX7is 17mm
The 5751 is very close in THD to the AY7 - both are low. The Mullard is higher in THD but still quite low compared to a bunch of other ax7 that I have.
Well I have my bigSEamp more less shaken down well good - I think any more improvements are likely to be hard to find.
My noise floor is now easy to reach -82.1dBu in actual usage, with the guitar input.
ie. guitar -> gain stage 1 -> EQ -> gain stage 2 -> virtual earth plate follower mixer -> make up gain stage -> driver -> finals
That's a whole heap of stuff, and the noise floor is the 'worst' with that chain. ie. max nbr of stages in circuit
Here's the spectra at the speaker terminals for the 15" 4ohm speaker - with the guitar turned down, and fully up.
There's a minute or two of historical trace in red.
That's around 15Wrms - pretty comfortable volume, with the THD very low - equivalent sine would be < 1% easy.
There's still plenty headroom for transient, random, overloads without getting gross THD. 8) 8)
This is basically where it is ready to use
It sounds crazy good. Certainly the most refined guitar sound I've ever pulled. 'Harmonics city'.
My noise floor is now easy to reach -82.1dBu in actual usage, with the guitar input.
ie. guitar -> gain stage 1 -> EQ -> gain stage 2 -> virtual earth plate follower mixer -> make up gain stage -> driver -> finals
That's a whole heap of stuff, and the noise floor is the 'worst' with that chain. ie. max nbr of stages in circuit
Here's the spectra at the speaker terminals for the 15" 4ohm speaker - with the guitar turned down, and fully up.
There's a minute or two of historical trace in red.
That's around 15Wrms - pretty comfortable volume, with the THD very low - equivalent sine would be < 1% easy.
There's still plenty headroom for transient, random, overloads without getting gross THD. 8) 8)
This is basically where it is ready to use
It sounds crazy good. Certainly the most refined guitar sound I've ever pulled. 'Harmonics city'.
Attachments
Here's the 'sine wave 1KHz' spectra for the same amp setup as above but with the 8ohm load box.
Approaching 85dB signal-hum margin at a modest input level of 1Vpp and a well low THD of 1%, nearly all H2.
I think if I used the second psu choke (I ended up taking it out - it gave around 1.2dB improvement to the bottom line), and did the whole thing again from scratch but *tidy -like*, I could almost reach a worst case of 90dBu hum floor - which would be 'the uber-perf' benchmark for this platform - at the speaker terms, flat out for 1% THD.
If I really need that perf level, I can just switch to direct mode (s)
Approaching 85dB signal-hum margin at a modest input level of 1Vpp and a well low THD of 1%, nearly all H2.
I think if I used the second psu choke (I ended up taking it out - it gave around 1.2dB improvement to the bottom line), and did the whole thing again from scratch but *tidy -like*, I could almost reach a worst case of 90dBu hum floor - which would be 'the uber-perf' benchmark for this platform - at the speaker terms, flat out for 1% THD.
If I really need that perf level, I can just switch to direct mode (s)
