REW Update

[Tubetec]

Its a bit like having thousands worth of hardware with a HP badge on it , but its all in software .

Maybe a hi-z input jack on an audio interface the would be more appropriate for direct to tube connection  with less loading effect .

Any good tutorials on REW out there ?

Is it possible to produce graphs of output level and distortion vs frequency ?

Looks like the program runs reliably on everything windows at least as far back as XP ,
Any old pc or laptop can be repurposed for this , even with a 2009 low powered single core mini pc  all the metering and general operation of the program is fast  and smooth , really this program deserves to run at all times on the test bench on its own dedicated little pc , what would be nice would be to take your idea Alex, with an attenuator box ,but add an ultra hi -z cathode follower input to  a 384khz capable A/D -USB , just like in the days of old the cathode follower input could be built into the probe itself , the USB housed with the psu in the enclosure .

It seems like the usefullness of the program in terms of electronic signal sources has barely even being explored on the site , a little  box with a hi-z probe and attenuator would make a great addition ,all of a sudden it becomes a serious test bench tool ,Once I explore it more its gonna be lots of fun tweaking my monitors on it ,but I think  its worthy of everyday bench use and not just speaker tweaking .

 

[MountCyanide]

Thanks for hipping me to this. Installed and ready to go.
And if anyone from here does want to do a video tutorial on REW <in Los Angeles> I'd be more than willing to shoot and edit.

Cyrus

 

[alexc]

Regarding measurement of high ac signals, like those at the plates of tubes etc ...

Because there is usually a large dc voltage present, it is safest to make up a box with a large dc blocking 600V poly cap - I use an orange drop type 0.22uF to  0.47 or so as I recall. 

It's required to prevent the dc at the 'test point' finding it's way to your expensive sound card inputs.

I have tried a simple attenuating CRO probe as the 'front end' for my attenuating box, but it's unsafe (crack! burned out/shorted) ..  small components easily fried with possible damage to the sound card.

I also like the seperation of a single-wire HV sampling part with the ground wiring in the box and breaking out on a test clip. It's best not to have the probe active tip any where near a ground (as they are on a conventional cro probe with the front hook removed)

One also has to remember to 'ground the probe tip' when not in use - to avoid unrequested discharge!

This is also the reason not to have to high uF in the dc blocking cap. 10uF electro is a no-no. That's a lot of discharge to go off.

As to accuracy, in most cases, the loading effect is a few percent; occasionally one needs to account for it but most often not.

Generally you can safely keep an eye on your noise floor as you go and prevent hum from getting away from you.

As far as using a HiZ input on a sound card for the probe - I do all my I/O balanced with XLR. I have various balance/debalance boxes and cables that I use as required.

It also pays to use an interface that is 'not too expensive' just in case you do somehow fatally spritz it    For same reason, I don't recommend using one's computer's built-in soundcard facilties for this job!

....

I can't imagine not using REW to nail down performance at my plates  ..  and all over the place ...  checking the hash on power supply lines, B+, B1+ etc, grounding schemes  .....  its all there. ;D

For me, there's always been some pathway found to a significant improvement by spending some REW time.

I do annotated snapshots at each baseline of change in 'all my circuits' ... it's a discipline that pays off in terms of making informed decisions about  performance and tradeoffs.

I mean, if you can't demonstrate it (with a REW snap),  then it doesn't  exist, right ? Especially for nebulous small dB changes and what not.

The danger (of over reliance on 1st level measurements) is -  it can shape ones 'paradigm' towards 'less THD' ..  some times I thought I had a 'sound' that was exciting but on subsequent measurement-and-tweak, I felt I lost that 'thing'.

Not often though!  Still haven't come across a case where I prefer more noise.

Character, sure thing!  Hum and such - no, no, no.

 

[alexc]

Here's a nice REW snap 

Its a tube build with a quad New York Dave One Bottle module  powered by a remote modular PSU with a heavy duty 7pin interconnect cable of 2m length using cheap and large-ish, eby 'aircraft' connectors.

Regulated HV and dc filtered heaters. The PSU is a 'twin' layout, each section is identical and independent with it's own interconnect to a signal rack.

The  PSU modules each do up to +350Vdc 100mA, +6.5Vdc 2.5A, +24V 1A, enough to do a decent signal rack or two.

Here is a REW snap for  one channel, with a nominal -20dBFS balanced send and using my 'hv ac probe' box to safely attenuate the return signal to suit my audio interface's balanced input range of around 0 .. 8Vpp.

Here, the probe box is doing something like 30dB of attenuation (on a rotary switch)  ie.  500Vpp divided down to 10Vpp or so

...

Fairly good performance from the nos Philco ...  minimum nfb setup  ...  57Vpp plate signal being measured  ....

ie.  a single av7 tube with 2 cascaded gain stages and nfb  ....    gain x57 

(driving a load consisting of a 15K/600 Edcor xsm output transformer)

0.4% thd+n, nearly all h2 with 87dB of signal-hum margin and +27dB of headroom    Hum floor is at -96.0dBu. 

When I crank the input signal some, I get a 1% thd+n scenario with  >96 dB signal-to-hum margin ...  at a plate doing 120Vpp signal into 15K load.

That's a hum margin within 'sighting distance' of  typical opamp circuits that have some  gain.

Not bad for a modest AV7. The nondescript 'cascade-with-nfb'  beats any single tube gain stage I've seen ...

...

So, all that's pretty good I think!    And around 100usd per channel including tubes, signal traffos and  very quiet PSU.

.....

I estimate the PSU is more than 3dB quieter than my best regulated 'co-located' psu to date. 

Something like 9dB quieter than the psu in my typical 'amp head with rectifier+choke psu'  type builds.

Can't wait to do the 'limiter' and 'eq'  modules     

.....

Modular, remote and regulated PSU  ...  rock the house big time for 'over the top' tube stuff.

[even more so if there are any signal traffos about]

Give yourself a 10dB noise break.    Go remote  (and modular !)

 

[alexc]

Here's a pic of the biz ..

 

[alexc]

and the NYD module  (from a great Group DIY pcb designer ! )

....

The modular psu also provides a dc rail for powering some class A solid state line amp modules - 'JLH Headphone Amp' modules.

Super cheap, small  with serious drive capability at figures hard to believe from just 2 small and 2 big transistors.

I'm using a heavy duty LT1083  regulator module within the signal rack to provide +24V at 1A, enough for 4 to 6 JLH line amp modules (biased for low current )  and needing quite a large, robust heatsink!

The idea is to get that beautiful NYD signature to line level without destroying either the wonderful harmonic structure or the exemplary hum margin.

When I rack it all up, it will be a small mixer ....  4 channels to stereo with direct outs and 2 independent pairs of stereo outs. 

I'm using unbalanced mix busses  with 4 sources and a tube sum amp  ....    'plate follower+gain' config driving output transformers. Once again, the modest av7 - they really are a nice performer, with a wide range of applicability.

The busses, like the input channels,  will be boosted with JLH line modules driving some 600:600 utc traffos.

It's a rack with 'tubes and transformers' inside  and  also on the 'top deck'    All that extra space, you can do with.

At a rock bottom price, of course  ;D

 

[alexc]

Yep - I surely measure flat-flat on my 'hv ac probe'box setup for the audio 20-20K

and even as far as around 48KHz or so before my signal chain and audio interface and uP horse power finish up.

And the low to mid  ratio matching transformers available, even at a budget, can measure super well for audio  .. re  Edcor XSM and similar. In static measures (at least) they do very well.

 

[trobbins]

Where do you think the 50Hz ingresses?

 

[alexc]

The 50Hz is the 'ripple' that the HV voltage regulator leaves  ...  around 1mVpp on 325Vdc  or around 0.3mVrms which shows up at -117dBFS on the spectra, up from -130dBFS 'ground hash' or background noise.

That's a pretty good figure as far as things go ... 

I am not quite sure why it's 50Hz and not the more expected 100Hz (post rectifier/resevoir caps). 

In this case, at least there is no appreciable  'induced' hum(from the psu traffo)  into the transformers and highZ parts of the signal unit ..  the psu transformers are way away.

Normally I get a certain amount of 50Hz, which is mostly 'induced' and a stronger 100Hz (+harmonics)  hum which is the B+ ripple.

I'll be monitoring it as I increase the loads on the psu ...  it's early days yet in the build.  But so far, so good.

 

[Tubetec]

Do the mains transformers you use have an interwinding screen ?

I see a lot of modern transformers dont have it , but its very common on older stuff .

I read somewhere that using very large caps right after the rectifier can be problematic also ,as it forces the transformer to supply very high short term peak currents ,and that in turn can cause unwanted effects and coupling into other windings .

There are a few older guitar amps ,which seperate psu/power output  and signal chassis , even though the physical seperation is small , it does help ,as strong AC currents make it no further than the psu .

Just got gifted an old Ac-30 choke ,that needed replacing ,due to total breakdown of the wire insulation , I managed to strip it down and keep the bobbin intact , so I might have a go at rewinding it , I notice better quality chokes often use paper interleave between layers , this particular late 60's vintage one had no insulation between layers,just a big ball of wire , it measured .75 HY and 26 ohms resistance , as I unwound it the the degraded wire was easy to see , should have measured 20 Hy and about 260 ohms or so .

 

[trobbins]

Winding wire insulation is just so much better than it was in vintage days. The need for inter layer insulation for a choke is likely rare. Main issue would be lead wire insulation and creepage between first and last layers, and to core.

 

[alexc]

Some interesting insights guys - thanks.

My psu traffos at the 'affordable' end of the Edcor XPW  series - no 'screen' between the windings, I think  .. and also no copper 'strap' around the outside either.

Originally I used them in little amp head circuits  ....  but I did find them to be quite noisy!  At the time I felt the radiated hum into my signal traffos,  a UTC A series as well as a Magnetic Components  '6v6 deluxe style' ,    was very, very noticeable.

Unusable like.

I've since confirmed that *both*  the little Edcor psu traffo is a little 'splattery' (wrt radiated fields)  AND the utc a series (4K:500)  are quite prone to hum with even a whiff of HV ac nearby ...

Hence the push to 'external psu' units on my current builds  ....  *and* the utc  a-33    shielding cans (such as they are)  PLUS  some extra  heroic efforts for them to work well in the 'deluxe' amp head rev4 (now my daily practice amp)

Thanks to REW I did find these irons'  particular orientations  *and*  physical seperations necessary  to get it all hum happy.

Distinct from the 'remoting' of the psu,  the major contribution to hum reduction in a small amp head context,  has been the shielding/shadow  effect of a larger-and-heavier-than-usual  6v6 output transformer ....  in between the utc signal traffo  and edcor psu traffo 

....  to get an instrument amp with hum-noise margin at the 'high end'  of >85dB in typical usage

................

Nowadays I go all round  ..  serious shielding on signal transformers as well as as remote PSUs ...  to give my more ambitious creations a big starting advantage  and a reason to exist.

I've been lucky to get quite a bunch, like 10 or so ...  of some local 50's broadcast traffos for fair prices - heavy duty shielding including the 'cast, nested cans' types, reminiscent of the glorious US made originals.

The typical 6v6 pp plate-to-speaker transformer in these lines are huge! compared to the 'deluxe' traffo!

....

The downside is you need to make a bunch of large racks - I make mine oversize for good  seperation between units as well as for good cooling, despite the use inexpensive solid pine boards.

More like tables and open-back book shelfs, really. The light stained timber really showcases the more elaborate 'tubes'n'traffos on top deck' style of construction.

Further on that, I'm doing up a pine rack that is 'long' rather than high .. and can also take my standard RU chassis-es  .. 

It's really liberating to just 'let it all hang out' regarding the user control panel, and let the layout organically spread out for 'comfort'  .... as opposed to making everything fit in RU modules  ..  often each with their own psu and mains traffo and all.   

Going for an 'rca radio console'  kind of look, with big rotary faders and VUs and such  ..  and an  angled workbench leading to a couple of conventional style tall racks.

I having a try with some brass and copper  metal plate for my 'mixer' section - a bit different from the more usual alu.

To augment the proper old school VUs, I'm installing a couple of cheap eby led meters  ..  the kind that do some peak-hold on an array of 50 leds or such.

 

[ruffrecords]

Talking of internal vs external power supplies, I found the biggest culprit was the 12V dc heater supply I used in my tube lunchbox builds. Big toroid followed by big bridge and reservoir cap followed by LDO on big heatsink. The toroid is asked for some serious current pulses by the rectifier. When fitted in a rack with four tube mic pres, the 50 and 100Hz hum level reduces as you move further away from the supply (REW was great for quickly comparing outputs). The electronics was in a screened box and I even invested in toroids with mu-metal screens which helped but hum was still affecting the noise floor (I think some was being picked up by the the unscreened Carnhill output transformers) . Almost in desperation I replaced the linear heater supply a Meanwell 12V SMPSU. BIngo, no more hum. REW  measured the best noise floor I have ever achieved with a tube mic pre.

Cheers

Ian

 

[alexc]

Thats interesting Ian - may I ask what is LDO  (I know now  ..  Low Drop Out regulator chips 

So the regulated dc heater is doing a fair amount of current at 12V and was contributing to hum  ... and the system  was quieter with a switching psu.

I guess it has to do with 'dirty current' return from the heater circuit getting into the audio ground ?

Or a case of the large current spikes spattering em fields which are being induced into wires/signal traffos ?

...

This is my first external psu and it can provide about 5A of dc filtered heater at +6.5V from both psu modules ...  no sign of anything so far hum wise from this source.

But I've only just started 'chaining' signal racks off of it.

I *did* notice a big difference with either 'ground referencing' the dc heater negative at the signal unit or at the psu unit or just floating.

I was trying to determine if heater current would contaminate audio ground  and if floating heater was useful.

I found the best by far was 'ground reference' the heater at the psu unit. Not much different was doing it in the signal unit. 

Really bad was 'floating' - again not sure why! It was a significant broad hash of noise across the hum region.

It could be that there is no 'return path' for the heater current ... ? Maybe it leaks into the cathodes to ground or something like that because there is nowhere for it to get 'to ground'.  I don't know.

The tube was amplifying fine  just a frightfull wideband hum.

One day, I also have to measure and compare :  ac heaters floating, ac heaters with centre-tap grounded, ac heaters with one leg grounded (as is common in amps), dc filtered floating, ground referenced and/or  regulated      ;D

....

For now, I will be keeping on eye out with REW as I load up the psu with more (and more) tube stuff,  then to also ding the 24V winding at 1A of JLH goodness  8)

....

Modular PSU #2 is my current attempt at a higher powered psu and has a much more serious set of psu traffos. It is intended to power some tube monitor amps on one side, and a mastering chain with major tube limiters  on the other side.

A 2RU with modules on the inside, toroid inside and big Edcor and filter caps on top. Again with 2 cables but this one having individual power and standby switches for each connected cable and it's associated psu traffo.

One psu traffo, the large Edcor,  will be providing some 6A of heater for the amp rack, of dc filtered with seperate windings for the finals and the preamp stages.

The other psu transformer  is a 100VA 2x 12V toroid for heater on two windings for seperate heaters amongst the mastering chain units.

I expect this PSU unit to be doing 11A of heater all up. The HV will be a combo of unregulated for the finals' some 250mA worth, and regulated for screens and preamp/gr amp sections able to do 150mA  ....

A serious psu but quite straight forward and hopefully ..  robust. So I'll be  REW-watching  as I go.

....

And a final bit of kit .. invaluable when racking these kinds of powerful amounts of heavy metal .....  one (or more)  standard rack 'tray' s  made from heavy duty steel plate  ....  4RU, open at the front, tapering to a 40mm high rim at the back which is quite deep, like 400mm or so  with a floor of same  ...  that nice old yellow coating on steel  ...  2.2mm or so thick.

They used to be cheap in the old days. Great EM shielding and does indeed make a diff to the noise bottom line  in the right circumstance.

You stick that above the psu #2 with plenty of space underneath it, on the underside of a shelf, then on top you start  with the amp module, leaving plenty of space above it too .. and so on.

...

At this point, with PSU #1 looking really good, I'm not sure if I can do any better - I suppose, some more improvement could be had but I can't imagine it more than a dB or two.

Theoretically, the hum could go down about 12dB more,  till it is buried in the noise floor;  which is -130dBFS in my measurement system.

- better HV regulation module ?    I get 1mVpp of hash with no sign of modulation at all on my $20 eby hv reg module.

It already uses a high speed error amp in transistors around a good quality power mosfet. The rectifiers are 4x hexfred diodes which have no 'snubbing' capacitors across each.  That would be my next thing to try.

- IEC filtered mains inlet connector ?  The next model I have a really heavy duty one of them, japan manu

- regulation of heaters ?  possibly, but I've never been a fan especially at any significant current. But is is possible to get beefy linear reg modules and also switched mode psu modules.  It is an area I'd like to try some more.

- seperation of modules ?  Already done

- better layout of signal modules ? Surely a plus

In general terms, with no 50Hz ac anywhere near the signal unit,  its more about high amplitude audio buzzing from one part of a circuit to another  .. from the output end into the front end and/or left-right channel seperation  etc. Not much impact on hum  at 'silence' really.

That and of course, grounding ...  always the grounding and the shielding. Return currents and seperation of etc. Shields and seperation from audio grounds, except of course, when they is not.

- better interconnect cable ? Possibly an even heavier ground line would help. I'm already using an interconnect with wires that are heavy gauge.

- better performing tubes ? No doubt!  Some tubes are better than other tubes.

There could well be a couple of dB still of better signal-hum margin for a specified THD to be had from 'tube selection'.  Rolling tubes is, after all, the 'last refuge of the simple'    That would be me, for sure.

I have a set of around 20 candidates to have a look at in this signal module ...  av7 and at7 types  ..  so far, it's a close spread, within a dB or so.

My next signal unit is the fun one ..  point-point wiring, with some octals as well as the smaller 9pin types  ...  so I can go to town and chisel on down,  past the 100dB margin of the old low THD tube hum  ..  and  into the great beyond.

There, I'll need cooling; like liquid nitrogen,  to keep out the cosmic rays  .. that might impinge on my grids.

......

I  haven't even looked at other forms of noise apart from 'hum' to date - there's also a world of 'switching transients' type of noise related to rectifiers, as well as  'hiss' type  noise and a good many other things so far not even on my personal radar! 

....

Anyway, even if 'that's it' regarding hum performance, with this psu #1, I'm happy to have REW'd (subdued) the tube beast even to this level!

Next up ...  JLH mini-modules all round       

....

Here's a snap of my 'opamp test rig' silence noisefloor ....  basically  a variable straight-wire gain with balancing/debalancing using some of the quietest chips around ... LM49860 for debalance, another for gain and a DRV134 balancer,  with an Alps potentiometer set for unity gain at the non-inverting opamp.

All that x2 for stereo. And with a groovy machined alu 'shield' for the gain opamp!

It's about 2.5dB  22.5dB  better at hum than the tube getup I showed above.  ** correction : 20dB measuring error!

[I confused the real-world figures with a 'scaled' figure ...  sorry! . Explained later in thread. ]

That opamp cct 100Hz spike is still about 10dB higher than the ground hash of -130dBFS  .... the test rig is powered by a 'JLM 3way' regulated linear psu at +/-16V.

The 25VA toroid that powers it is 'co-located' in same 1RU but this isn't much of an issue for builds without signal transformers, and using low voltage rails and lowZ chip circuits.

I have some 'toroid shielding strips' that I use here and there ...  in this case they make a tiny difference, some 1dB or so I recall ..  once the toroid has been 'rotated' and so on for min hum.

TheyJLM 3way modules work well, but  simple lm317/337 regs are not the quietest thing available, in terms of ripple and so on. And they're best suited for the lower current requirements even at these low voltages.

I since have acquired some psu modules, similar in size to the jlm that I've generally used up to now - these are cheaper and have discrete voltage regulation circuits - they claim to be very quiet, so I'll be checking that too.

For Class A circuits especially, a really low noise supply is much better - in small current apps, little discrete 'Class A regulator' circuits should do really well.

For the bigger jobs, they can scale up or one can go to LT1083, LT1084 based linear regulators which I believe have a tighter performance spec as well as much increased power ratings.

I have one module that does +24V at 5A easily with large heatsinking - for those intermediate level JLH and Neve styled requirements  ;D

....

I did some REW tests of my dual channel gyraf pultec eq .... bye-bye to the tube psu and make up gain module!

So much ripple, hum and hash  from the co-located psu  ... ... when boosting the bass freqs 

I tested the unit  after the removal of the psu + make up amp modules  ...  using this opamp test rig  ....  the filter results were really good and the eq sounded totally clean and quiet  with the opamps doing the make up and line amp duties.

I'm keeping the 600/600 input traffos and adding some more at the output end.

It was a blow tho' - those Lundahls in those modules cost a lot and aren't much use without a pcb

Give me an old school traffo any day - you can recycle, and recycle and on and on no probs.

Just finishing up some 'ssl mic pre' pcb modules now to do that peq makeup+line  job and finally realise the full potential of the thing, after 10 years!

 

[ruffrecords]

My problem was definitely induced because I remoted the supply and the hum disappeared.

I always reference the heaters to analogue only at the power supply. This is unavoidable for me because the heaters are elevated relative to analogue 0V.

If you float heaters you will always get hum. I am not entirely sure why but I have accidentally done it enough times to check it first if I get a lot of hum. Anyway it is not good for the tubes. They have a specified maximum heater to cathode voltage and resistance which floating most likely violates.

For the Mark 3 tube mixer I am going to attempt 100% SMPS for all supplies. I'll let you know how I get on.

Cheers

Ian

 

[alexc]

Thanks Ian.

I wish I was just starting out with the benefit of all the great insights and knowledge around this place.

 

[alexc]

And to wrap up my commentary, here is a proper 'old school' piece of kit I just acquired for fair price ..

Genuine way-er  'down under' ...  affordable  ...  destined for front and centre of my test bench  8) 

This is how they rolled circa 1955  .. measurement amp style!

 

[alexc]

and an angled front pic  ..

 

[alexc]

.. inner

 

[alexc]

and oblique rhomboid ..

I'll refresh it  and just maybe, a few mods to the ins and outs .. a new 'probe-ulator' perhaps.

Lots of REWing for this one!