Measurement Software - 'Room EQ Wizard'

[alexc]

ps - to be fair to my aging ears, I generally used these at levels less than max power  ;D

More like 10W peak or so, well below the 1% THD level (before optimisation)

Still - big THANKS to Room EQ Wizard for the wonderful THD measurement capability!

Think I'll have a listen - I'm curious what the sound is like with no nfb  : 8) 8)

 

[alexc]

Oh yeah ...  that's sounding better. The tops are much more natural now and the need to back off the highs at 3Khz has much reduced. I have flat eq on the speaker and it sounds quite balanced and powerful bass - seems to be more solid in the lows.


Fantastic. Hum is non existent, even at full power and ear near to speaker! It is actually better now (with the PI improvements) without nfb than it was with nfb. 

I have the other mono-block with no improvements here also, so I can do a side-by-side A-B listen.

 

[alexc]

Back onto optimisation of VA stage, currently a 6N6P tube.

It's an interesting choice I made there, possibly misguided!

Because this amp doesn't need a lot of gain, being driven  by a hot Finalizer signal, minimally attenuated,  (around 6dB attenuation on the 'Level' control at 'normal' listerning levels  , I end up with typical inputs at the VA grid of around 1.0Vrms to 2.5Vrms.

That's around 8Vpp on the strongest passages - meaning at least +5V or so bias (or, of course pad the grid signal some!).

So, I wanted  a triode that can be biased around  +5V to +6V and with low rp and low mu and an amp with generally low gain and low nfb.

Those things go together, to a certain extent, and indicated an AU7 class of toob for the VA.

Did the calcs, it 'leaned' towards more of a 12BH7  A bit beyond, actually. 

The 6N6P in there at present is something of a 12BH7 on moderate doses of steroid.

It has two triodes of 4W each, and a heater current that is a bit higher at 750mA - has a very low rp at 1800ohms and the capacitance grid-anode is at 3.5pf, which is middle of the pack for this class of tube.

And it has a decent max Vplate of 300V.

That means quite a lot of plate current (some would say a crazy amount!)

It needs to only provide 25Vpp output at low distortion and with a not too bad PSRR, given that it's lightly plated loaded in 8K6 resistive.

I'm using a single stage only (don't want to increase input capacitance)  and it certainly is doing the job;  direct-coupled (no coupling cap) to the PI grids and I have elevated the 6.35Vac heaters by 70Vdc.

I measure around 0.85% THD+N for that stage at present. Which is still too high. And, the relatively high Cg-a may well be contributing to the moderate square-wave ring I observe.

SO, imma press on and rework the VA with a somewhat less crazy tube - I'll use the extra gain for increasing the nfb later.

My choice here, is to go for lower Cg-a of 1.5pF and a lower current (but still quite a bit!)

I like the 5814A in this revised role - the downside is a significantly higher rp of  7700+ ohms at higher plate voltages.

I'll be using it with around 14K load, at 165V and around 9mA with a bias around +5V

Looking for lower THD firstly, better square wave (no global nfb yet)

Step 1 - use a single triode as a grounded-cathode gain stage
Step 2 - add the second stage as a cathode-follower
Step 3 - change second stage to parallel

Then I'll review  the data  before going further. 

My guess is I'll wind up with the CF tacked on
I always seem to want to do that (even before I had quant. thd data).Especially with an AY7.

As soon as I move away from an AU7 I end up with an AY7 gain+CF setup So far, always the EH.
But the nos bug is biting again and I'm wanting badly to do a 5814A thing.

----

The good news story is at least I am in the ballpark now, at full power.
I have the other monobloc for comparison and I've got time while I e-wait for new resistors and caps to show up.

The power of the THD meter compels me!

 

[alexc]

A few more hours of auditioning, EQ'ing and A-Bing ....

This 'A' bloc stays as it is, for now. Raw, with 6N6P but no nfb, but running comparatively well.
Compared to last weekend, that is! Curse my wooden ears!

The 'B' bloc gets the 'new' treatment with the AV7 and the nfb, already,  as well as the  'tune up'  8)

----

HOWEVER, as it is now, it does sound good. Very good, now that I've re-co-jigered my finalizer to better  'work the room'.

Which is all wood and  rectangular! Not optimal, to say the least. It's like a speaker box .. within .. a speaker box 

Where it will all end, who knows? It depends on how much  nfb sticks,  going forward.

Up till REW ! I didn't really has a clue!

Just ears and qualitative THD for most of the time. Occasionally I'd do the manual calc, then go and have a sleep

For now, this ChnA stays the same - a raw, but optimised version of last Tuesday.

Tomorrow I'll put in  an AV7 and some nfb in Chn 'B'   

---

ps - the finalizer is doing all the switching (digital, analog) and for analog inputs I currently have going :

- eq
- dynamic eq
- mb compressor

eq does the lo-boost and mid-high-cut  with some 20Hz lpf and some 3dB air at the top end

d_eq does a simple containment of the hi end 'flare', which is considerable and gives me an equal measure of the scowly face 

mb_comp does more '3 band dynamic range' reducing -  beyond a threshold, it cuts the top some more and does some balancing of the tops with the low-mid and a  much bigger bottom end.

---   

It really is vital for getting an 'engaging' sound  beyond around 10Wrms levels 

And making me remember what good sound sounded like (to me) back in the seventies and eighties.
Which was sometimes, very good.

Even so, when this exercise is completed, imma treating my lounge room to a sensibly priced sub  ;D

 

[alexc]

While I reconfigure my amp under test ....

- here is a Room EQ Qizard  THD plot for my bench interface Motu 828 Mk2 with balanced analog out looped back into the front panel mic pre #2

The mic pre has +37dB of gain  in this plot, and is sending -3dBFS to the asio audio susbsystem - which corresponds to -18dBu in the world I inhabit.

THD is 0.0022% and THD+N is 0.0398%   

not too bad, really and for my purposes, good enough for use with the budget price but well spec'd Dayton Audio measurement microphone that I will be using.

 

[alexc]

The Dayton EMM-6 tech info is http://www.daytonaudio.com/media/resources/390-801--dayton-audio-emm-6-quick-reference-guide.pdf

Here's a summary :

Capsule type: 6 mm electret condenser
Polar pattern: omnidirectional
Frequency response: 18 Hz - 20 kHz
Impedance: 200 ohms between pins 2 and 3
Sensitivity at 1 KHz into 1K ohm: 10mV/Pa (-40dBV, re. 0dB = 1V/Pa)
Max. SPL for 1% THD @ 1000 Hz: 127 dB
S/N ratio: 70 dB A-weighted
Connector: gold plated XLR
Phantom power: +15 V to +48 V
Weight: 144 grams

Should be fun - if I can get useful info and get me to the next level of audio performance, I'll be happy.

 

[alexc]

Still playing with Room EQ Wizard

Here is a transfer function of the 'measurement' signal channel represented as 'spl'  (magnitude in dBFS) and 'phase' (in degrees)

The 'reference' channel is simply a balanced xlr-xlr cable from Motu analog output to analog input.
The'measurement' channel is a Jensen traffo DI box returning to the onboard mic pre and some gain.

Again, I think it's believable - the magnitude response is spot on with my 'standard' TrueRTA and I think I've described the basic calibration of my audio system to dBu reference. To a certain 'confidence' level, that is

The phase data is relatively new data for me. I usually use the cro for that, and I will check it further. But for now, I have good confidence in the plot.

Again, all very easy.

 

[alexc]

Here are the THD plots for the          Motu Analog Out -> Jensen DI -> Motu Mic Pre  channel.

Awesome measurement cursor table at the bottom, which updates on a mouse click all the traces on the measurement cursor position. 

Just great. You can easily see the THD+N peak at 20Hz.

ps - I have still to calibrate for Room EQ Wizard.  So far it reports on -10dBFS before runnng the automated tests.

 

[alexc]

Here's an interesting one ...  spectrogram of the Jensen+MicPre  measurement channel  magnitude  co-plot with the reference signal.

Zoomed in to max available on the plot, the time scale is 100us and there are (mostly) 4 samples per division.

So the observable inter-sample period is 25us, which is as expected for the current 44.1KHz sample rate.

Now that's pretty darn good - and it's relatively easy to get there on the plot.

WHEN I test all this at 96KHz, it will be interesting. Very interesting indeed.

ps - one of my probs when testing my fairchild 660 limiter  'interpretation', was getting sufficient resolution from my 'trace recorder'.  It supposedly has an attack in the low microsecond (us) range.

I don't have a storage facility on my analog cro (digitally controlled), so  no 'low microsecond  waveform sample history' there that could be used to verify what attack period I actually am getting.

So of course, I used my DAW to record the limiter response. I did it at 44.1KHz, and can see the sample-sample trace but of course at 44.1KHz, the limiter action leading edge falls between samples.

I measured it as being 'in the ballpark' of <50us or so and left it at that. It was something of a low priority to redo the tests at 96KHz and I had other things to ponder  ;D So there it remains.

Shortly, I'll recharacterise that pmSix60 limiter with all this renewed interest and toolage!

 

[alexc]

Just looking at the Impulse Response - it's all in depth and quite amazing really.

Really worth learning how to drive and interpret, particularly in the speaker, crossover, damping factor, network impedance, group delay and tonnes more stuff too.

It's like being at university EE, but with great lab tools of todays world not that of 30+yrs ago, when I first started with all this  8)

And a whole lot less cost!

 

[alexc]

Back to the amp - I did the change to a nos rca AV7 and did a rough setup.

Main amp now has a no nfb THD+N of 1.0% at 15.5Vrms into 8ohm load box, 20-20KHz measurement bandwidth.

That is a big improvement!  I'm e-waiting now for the resistors, but I think I can get that down to around  0.85% when I get the right values setup and re-align everything (again)


The gain has increased, as expected with the mu of 42, and by around 6.5dB. That and the slightly improved gain from the PI changes  gives me around 10dB for nfb to finish the rejigging.

Target full power THD+N is now  < 0.8% 

 

[alexc]

While I wait for parts, some more Room EQ Wizard work :

- calibration
- use of impedance sensing resistor and calibration for impedance measurements 

 

[alexc]

Today was a buy for some more nos av7s - including a few of the ge sub 6829, and some 5814a for good measure  ;D

Will be auditioning and measuring those in the 'new look el34 amps' in a couple of weeks

....

And while I get to that, I couldn't resist scooping up this big fella and getting ready to do some testin' ....

Its my build of a big Edcor SE amp, with monster Hammond psu and a pretty-darn-huge Edcor hv choke.

On a chassis from a real nice us. gent, who I believe is passed on.
But he did 'good work' - what higher praise for any audio related guy?

Anyway, finals are currently a pair of early seventies rca black plate 6L6s, but it can take pretty much anything octal - has individual bias for each of the parallel'd power tubes.

Lots of current behind that, Hammond 372JX 'export' primaries

  600V C.T. @ 288mA,  505V C.T.,  5V@ 4A, 6.3V C.T. @ 8A

I've used it up to 220mA on the Edcor no probs - it's a giant whack of power and with the 15" open-back I used it with, it had quite a bit of chest impact!

Now it's running at a laggardly 130mA thru the monster Edcor - quite relaxed and very forgiving of setup.

At present, it's set up as a guitar amp, but with a main input mixed with 2x aux inputs. Fender style at the preamp.
However, I've measured the response and its very flat 20-20KHz, and the distortion .. well, it rises with level!

The rest of the tube line up is AY7 on the front end main channel, AU7 for eq makeup and CF output, 6N5P to mix the auxilliaries,  a 6SL7 to mix the mains with the auxes, and a CF output to drive the finals.  Cool lineup, I thinks.

I am running ultralinear at a modest HV - something like 360V B+,  from memory. Well within limits for the old RCAs in there at present.  My first choice was some current issue tungsol 6550 - like you do, right? Run a pair in parallel SE for 30Wrms or so. into 16ohms. That's what I did - with nfb on a pot on the top deck.

I use around 10dB of nfb here, I think. Enough to squash the hum but not so much as to clean things up too much. It's already running in ultralinear,  it certainly can do clean and full BW. I have it set so that there is still some 'grind' at reasonable power levels in practice.

ie. at 15W or so with a high sensistivity Jensen 15" 50W max.

Big sound, on bass awesome.

Guitar can has the best of the 'champ' thing but at higher power and with a big bottom end. Pedal steel is great - with some pedals it has the body to be really engaging and not a ear shrrrredddeeer (like a Twin!).

SO, tomorrow I'll be running it thru the Room EQ Wizard analysis suite (part of!) and give it a good checking over.

As Kate Blanchett said in that last Indiana Jones movie ..  'I vant to know ...'

 

[alexc]

Oh yes, this one has a bal lined out with it's own 6U8A tube and an edcor xsm and a sensitivity switch.

and the back .. an aux , with another on front, each has it's volume control. Seperate  'mix' control on the front combines  auxes with the main channel - it has an simple hi-lo eq (which is just on the main channel).

So - lots to test for here - apart from the usual measurements of gain, hum, freq response sweeps, square wave tests, and such, I'll be focussing on the THD at each stage thru the amp, at various levels and various nfb settings.

I'll be very interested in the thd figures and what not - then I can critique my choices and prep for a 'Mk2' version.

While I wait for parts, that is!

 

[alexc]

Where I left off,

I have the 'probulator' (4.7uF 450V cap in parallel with  2M2 to amp-ground,  and behind that, a balanced attenuation network adding up to 1.1M. The 'sense' resistor is 82K of that. So, there is an attenuation of 25 approx.

ie. my 10Vpp max Motu range becomes 250Vpp, and the probe has a fairly high 1M impedance.

And the load box is similar - 8ohm 50W (5%) alu with a balanced  attenutation network across it, adding up to 60K, with the 'sense' resistor of 10K, so that gives me an attenuation of 6 or so. My range to the Motu balanced in is thus  60Vpp more or less.
The loadbox looks like a 10K resistor across it's hot/cold.

Now, scaling the large AC voltages is fine for relative measurements. To what extent the scaling affects the THD measures, I don't really know. It appears to be very little, and shouldn't alter as compared to the un-scaled (LV) measures.

In the future I'll do some testing and form a view on that issue.

-----

Of course it should be noted - stay safe with tube voltages!!!!

I use the probe with insulated clips which I setup with the amp powered off, then when everything's ready, I power on. Hands safely away.

Same deal when moving to another to test point - and I discharge the probe cap before anything!
Failure to respect 'hv probing' good practice can be very, very bad for your continued longevity!!

OK - on to the interesting stuff!

here's an inside pic

 

[alexc]

Looking at the VA amp first - probably my fav config here.

Its is an EH 12AY7 with one half as a grounded cathode stage with 68K on the plate and +2.0V bias -  direct coupled to the other half set up as a cathode follower. The output of the CF is cap coupled to a tone stack.

B+ here is around +250V, Current in each half is around the same at 1.7mA, with the voltage drop across the GCA  plate resistor about the same as across the CF cathode resistor (82K).

It drives a tone stack thru a big 0.22uF orange drop cap.

Gain is x27.7 or 28.9dB, so for an input of 3Vpp max, gives an output at the GCA of 83Vpp. The CG drops this a little, to 79Vpp.

....

So, with 1KHz sine applied to the amp front panel main input set for a 3Vpp signal to the grid :

REW signal generator, -12.9dBFS level output from Motu balanced analog out

GCA stage
--------------

AY7 (1) plate has THD  2.46%,  THD+N 2.46%  at 76Vpp output

H2 -> 1.64%, H3 -> 1.68%,  H4,5 -> 0.5% each, higher approaching 0

1% THD  35.4Vpp,  3% THD 79.1Vpp

response at 76Vpp output is    20Hz -0.4dB,  20KHz -1.3dB

CF stage
-----------

AY7(2) cathode  has THD  2.50%,  THD+N 2.50%  at 75.3Vpp output

H2 -> 1.54%, H3 -> 1.81%,  H4,5  ->  0.5% each, higher  approaching 0

1% THD  33.1Vpp,  3% THD 78.8Vpp

response at  75.3Vpp output is    20Hz -0.4dB,  20KHz -1.4dB

-----

So the AY7 here, in a moderate gain setup, has a reasonably high THD at 2.5%, and about equal H2, H3. components.
There is still significant H4, H5 dropping towards 0 for higher orders.

Not much noise, the THD being about the same as  THD+N; so supply decoupling is good despite the lowish CMRR of this stage.
(class A with 68K plate, no nfb)

The CF adds very little to that - around 0.05% and virtually the same freq response.

------

That's what I would expect from this stage, more or less. It's tending towards the high side of my expected range.

Low THD by my reckoning  would be around 0.7%, high around 3.5%

while I'm here, I'll do a sub of other EH AY7 I have here and also maybe an AV7 in that position for comparison.

----

Testing EH AY7 nbr 2 tube, for the GCA+CF  combo shows THD+N  3.34% at 72.7Vpp.

Thats quite a bit less output and quite a bit higher THD! Good thing that it is currently in a 6L6 deluxe type amp, as the PI.
There it tamed the whole amp down from the dangerously strident levels of the Groove Tubes AX7 it replaced - not as much gain, not as much 'grrrrr'  but .. sweeter at the chimes and more responsive  But still able to give more with the turn of the Master vol

So, out with that one!

nbr 3 tube shows very similar as last THD+N  3.6% and 74Vpp output. 

That one is used in rack preamp mostly the same as this SE amp. It goes back!

----

So for a sample size of 3 ehAY7, I get a range of 2.45% to 3.7% - to get the lowest would be a 1/3 chance. I do have some more of these in-use, which I'll measure another time.

How about some AV7 - here I have a sleave of  unused nos rca (70s) av7s. I tested the first one in the el34pp amps earlier  and it gave something like 2.5% in that circuit, which is just a GCA, no CF.

This  circuit isn't optimised for an av7, but anyway, as a relative measure and to check the spread of thd, it is a useful test.
These aren't burned in and I'm not letting much time to settle - just quick in and out.

av7 nbr 1  3.37% at 66.8Vpp output
av7 nbr 2  3.34% at 66.8Vpp   
av7 nbr 3  5.48% at 64.6Vpp
av7 nbr 4  3.31% at 51.6Vpp
av7 nbr 5  6.37% at 60.8Vpp
av7 nbr 6  8.20% at 60.9Vpp

and nbr 1 again for a reality check!  3.32% at 67.2Vpp

THD rises with warming ... then stabilises after 3mins or so.

------

SO quite a spread there on the AV7s in this 'quick-test'  3.34% .. 8.2%, with nbrs 1,2 a good match, 5,6 match but with more than double the distortion and 10% down on the plate volts achieved. Nbr 4 is a freak outrider , destined to sit in my little Champ (esque) 6V6 'character piece'.  nbr 5 is pretty ordinary and again, suitable for less demanding tasks

But first! I await the arrival of a set of 4x 6829 equivalents. If any of them match up with lowest THD, I'll use them instead.

A better test obviously would be to measure the halves seperately. That would characterise the tubes more accurately but not so the current configuration.

I'm not going  rewire to take advantage of uber-matching possibilities here,  because it's *this  amp*  I'm tuning, not a 'tube-tester'.

The first two nbrs 1,2 - lowest THD,  would be the best candidates in the el34pp  hifi amps - which is where they will go.
I'm pretty sure they were lower THD in that single triode GCA setup, at 2.5% or so.

However, the golden ear types do say that that's only a starting point!  The 'sound' is the main acceptance criteria.

I am not in that league - I can tell the diff between a AY7 and a AX7 though - and that's taken a long many years of guitars, amps and building!

----

Next up! 

EH AT7 - I use this in a limiter build, the d-oac.  6.23% at 95.9Vpp. Pretty high, but sample size = 1 !

EH AU7 - I use this all over the place for line stages, CFs and things that need 7.5mA or so. THD+N 2.21%  at 38.6Vpp

I have a bunch of nos 5814A and 5963 on-route - these au7 equivalents will be really interesting to check against the au7 types I have used up to date - mostly the eh variety and a nice bunch of Baldwin organ pulls from many years back.


And an AX7  phew 19.1%  THD at 72.6Vpp 

Note this isn't a fair test - the 12AX7 is not set up here, But there you are.

Dropping it's grid amplitude doesn't help much - about 12% is the minimum before it all goes bad. This is a really old one pulled from an mid-60s japan EL84 pp hifi amp.

AX7 nbr 2, a Sovtek of the cheapest variety,  shows 15.8% for 75.6Vpp dropping down to 7.5% or so with much reduced applied signal.

More on AX7s another time.

-----

Back to the AY7 nbr 1 - this is the original tube in this amp. 

Re-measure, THD+N 2.64%  at 76.7Vpp  (which is about the same as I found before)

-----

SO what do I conclude here from this simple first set of tests  ?

- firstly, the eh AY7 is generally a reasonably low distortion tube, and selecting from 1 in 3 gives one 'lowest' THD tube
  So far it's my best performer

- secondly there is a lot of spread in nos rca av7s
  I can get one good matched pair and 2 good singles, with one best suited in a guitar amp.

and , to get more accurate results :

-  the circuit should be biased properly for the toob-under-test
- should characterise each triode seperately for twin triode bottles

and finally,

- y m m v

----

Even given the rough and uneducated way which I am profiling my toobs ....

It's great to be able to make even these characterisations.

Room EQ Wizard's  THD measurement capabilities are wonderfully easy to use for these repetitious tests.

Tedious to perform but wonderful to peruse the executive brief!

I've only done this for the GR tubes in my limiters before. Now all future builds will be tested in this way and then have tubes selected to suit the purpose.

A great new day in my builds has begun

 

[alexc]

For this big SE amp, the original eh AY7 choice is my best option for now. My Super Reverb build has one in the clean channel first position  too.

I am to get a few nos 6072 in the hope of finding a good one for my c12 style mic. It too currently has the eh AY7  

I am a fan, you might say. But it's good to (so far) measure good performance from them (small sample size still, of course!)

----

Next up, 6N5P set up as a 'common plate mixer' - aux1 and aux 2 are each fed to a grid and the plates are summed and fed to a big orange drop 0.22uF cap.

So here's a good chance to look at triode 1 vs triode 2 in an identical set up.

No particular choice for the 6N5P here; I just have a bunch.  All I needed here was a fairly low sensitivity and gain. I plug pedals into these aux inputs, and I have a 1M log pot coupled to 0.22uF and then the grid, each side.

A bias voltage around  3.5 gives me a 7Vpp more or less as a max grid signal. 
So these can take a lot of signal before grid limiting.

27K plate loads for each  for a 3.5mA plate current at +150V off a B+ of 235V or so.  All pretty comfy.

CMRR is not great here (class a and low-ish plate resistor load) and so a well filtered supply is needed with good supply decoupling.

Which I have

tbc ..

 

[alexc]

Hold the phone!

Just tested the eh 12ay7 from my C12 style mic build - not so great.
THD 6.8% at 67.8Vpp output    -> output is poor, THD is v. high at more than double.

Reality check on best eh ay7nbr1 -> 2.53% at 77.7Vpp.

So eh ay7    is  1 from 4 for a 'best',      2 from 4 are 'ordinary' and 1 from 4  is a  'clanger'.

Not so different a distribution from the nos av7, but with an overall lower THD and higher output - as would be expected from a higher mu, more sensitive tube.

Looks like I will put the AY7 from this amp, my best so far, into the mic. Good news for the mic, not so for the amp.

The c12 type build, from the alctron, is set up with the with the same GCA+CF combo of the big SE amp I'm testing - a bit less current though.

- the so-called 'constant current CF'  GCA+CF combo relates to having similar current draw from the stage on the positive and negative signals by matching plate and cathode currents so as to present a more 'balanced' load on the psu.

-----

Just to expand on what higher THD means - the 'worst' eh ay7 has a THD similar to the 'best' but at a lower output level

ie. lowering the 'worst' eh ay7  grid signal  by 5dB, resulting in a output Vpp yields a THD+N of 1.5%.

So if you can live with 5dB less output, then you can have lower THD.

In the case of a mic, the grid signal is of course much lower, so that's a reasonable bet.
In the case of a 'tube test', then the closer the test approximates the intended app, the more useful the result.


In any case, I need to wait for more funds before delving into nos 6072 proper .

 

[Whoops]

Room EQ Wizard is a great software,
it also amazing that such a great tool is free.
I think thats the reason some people  ditch it and dont understand its full potential.

I have a lot in my hands at the moment, but it's great that you are using it and I will come back to this thread in the future to learn also how to use it in my builds and repairs

Thank you

 

[alexc]

Cool

Now that I have put my best eh ay7 in the mic, I'm I have my next best one in this amp.

Apart from the higher THD, I notice this nb2 eh ay7 the H2/H3 ratio has shifted from 50-50 to nearly all H3.
That could be seen as a sign of a questionable tube. I see how that progresses.

I think 50-50 is a good bet in this first stage, particularly in an SE amp which doesn't have the H2 cancellation in the output traffo. Each has it's merit, I prefer H2 but I think in this amp, later stages will accentuate that harmonic.

We'll see