the Poor Man 660 support thread

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
[silent:arts] said:
Expensive, and not available anymore.

On the contrary, with the schematic in the public it's now more available than ever! Replace the unnecessarily complex PSU with something simple and you have a great heavy duty "lofi" point to point project. In my opinion that's the way it should have always been.
 
Kingston said:
[silent:arts] said:
Expensive, and not available anymore.

On the contrary, with the schematic in the public it's now more available than ever! Replace the unnecessarily complex PSU with something simple and you have a great heavy duty "lofi" point to point project. In my opinion that's the way it should have always been.

If you use good transformers, spend some time to figure out a good TC setup, keep your HV wires carefully separated and match your tubes really well, it is far from being "lofi".

p2p? Why? there's nothing wrong with the excisting pcb's. Maybe the PSU needs a few tweaks but that's it. It's important to have a heavy power trafo, though.

Besides, why being so rude to someone asking help here? Isn't that the point of this forum?
 
telefunk said:
it is far from being "lofi".

did you ever put it through RMAA or oscilloscope, or send full range audio through it? It might be fine for tracking, but the base configuration not only oscillates easily, but also thumps in a really unique way due the equally unique cathode biasing. Even with perfect tube match it has the most peculiar moving headroom that really needs to be seen on the scope screen to be truly "appreciated". In other words it's an unstable fart machine, something people mistakenly associate with words like vintage. I hear those things are popular in some circles. Sure it's "color" but...  ???

It can be fixed - non-trivial - and dremel makes ugly PCB's. I'll compile a revision schematic one day.

telefunk said:
Besides, why being so rude to someone asking help here? Isn't that the point of this forum?

I can sometimes hide my contempt for the kit builders littering this place. People have different ideas of the point of this forum.
 
Kingston said:
telefunk said:
it is far from being "lofi".

did you ever put it through RMAA or oscilloscope, or send full range audio through it? It might be fine for tracking, but the base configuration not only oscillates easily, but also thumps in a really unique way due the equally unique cathode biasing. Even with perfect tube match it has the most peculiar moving headroom that really needs to be seen on the scope screen to be truly "appreciated". In other words it's an unstable fart machine, something people mistakenly associate with words like vintage. I hear those things are popular in some circles. Sure it's "color" but...  ???

We did check it out with scope, no oscillation for sure, also the freq resp was great, dont remember details anymore.

What do you mean "moving" headroom?

Calling it unstable fart machine is really weird, because my pm670 is far from color machine. It's as natural sounding as my LA2A or 1176rev.D or GSSL etc. Actually more natural sounding than any 1176 i've ever heard. Well, some might call those color boxes too...

Are you sure you didn't have some unsolved problems in yours? Also, in the "pimp" thread you state that your pm670 sounds "absolutely gorgeous". What made you change your mind?

Maybe you should save your contempt for something else as many of the kit builders gather more experience they will eventually start doing more creative projects etc.

 
telefunk said:
We did check it out with scope, no oscillation for sure, also the freq resp was great, dont remember details anymore.

What do you mean "moving" headroom?

Freq response depends only on transformers. It's impossible to screw that up. But IMD is more than 2%. Can be 4% (!) with this moving headroom, but that's difficult to measure exactly. It's of the "tube warmth" type I suspect you're not registering it with ears somehow, but it doesn't make it acceptable. Even a pedestrian vari-mu is not supposed to do that, but one with two matched gain reduction tubes should perform mastering grade.

The oscillation is higher than what my 100mhz scope sees and hidden in the sidechain, but it's there. If you have inaccurate gain reduction metering you won't see it either, but on decent ones it makes the measured reduction response wobbly and in stereo the two channels will become interactive on high reduction. The moving headroom thing can be only seen with scope with live audio (no idea how to make a video of it), but try feeding it the unit with a low level 100hz sine and then quickly boost its level: "ffbbraaabbrtb!". It settles in less than a second, but that kick was ruined.

I could post a revision schematic with data from all of the above, but like I said, this painting by the numbers building with has me aggravated. Why would I do that when some random kit seller and design stealer will come in and reap the rewards.

telefunk said:
Are you sure you didn't have some unsolved problems in yours? Also, in the "pimp" thread you state that your pm670 sounds "absolutely gorgeous". What made you change your mind?

perhaps link to a thread you're quoting. Last time I remember talking about "gorgeous" pm670 was maybe 4 years a go? Anything sounds "incredible" after €1000, whether justified or not. Yeah a G9 has "awesome bass".  Used to post about "ecstatic sound differences" in opamps too.  ::)

Then I started measuring things. Yes I did have unsolved problems in the unit, so does everyone else. Since then I have rebuilt the whole thing and PSU is my own. Fixed the instabilities, reworked the moving headroom issue.
 
Kingston said:
Why would I do that when some random kit seller and design stealer will come in and reap the rewards.

Because this forum is all about give and take and i believe it will, in the long run, be in everybody's interest to share the knowledge. Remember, like it or not, this is Analag's design and now that it is finally public, there is a chance that all the bugs can be solved and this might actually become a great PM670 MK2. Almost all the info is already there, it just has to be collected together.

But you do as you please, of course.

PS. I tried what you suggested and no "ffbbraaabbrtb!", no matter how much I tried. Why? It works with kick really well...



 
telefunk said:
PS. I tried what you suggested and no "ffbbraaabbrtb!", no matter how much I tried. Why? It works with kick really well...

No it doesn't! It's a stomp box effect. Look here.

Healthy level 50-200hz sine, scoped at the output transformer secondary, voltages in these images can be replicated any time. Yellow waveform is a reference of a perfect sine. These results are of course equal if measured on primary, but voltage 4X. In these measurements sidechain is disconnected and gain reduction stage is working as a static gain stage instead. (this is just for observation purposes, eliminating other sources of error, it doesn't actually matter if sidechain is connected or not)

Starting point:
latch-m670start.png


Quickly turn volume up, this is what should happen:
latch-m670good.png


What actually happens:
latch-m670fail.png


It settles into a perfect sine in about 500ms. Again, sidechain is disabled, this gain stage should be static, any distortion is very bad.

I wasn't perfectly accurate in explaining this "moving headroom" thing, because as you can see only about a quarter of the waveform is affected. So here's that same waveform distortion happening at much lower level (display set to 0.5V compared to 1V of the other images. That lower noisy part of waveform is a scope screen update glitch.)
In other words, this happens regardless of input/output levels.
latch-m670fail-low.png


The above is not related to transformer selection and happens on Edcor XSM10k/600 and Lundahl LL1676 alike. It happens on 6BC8/6BZ8 and 6N5P whether they are matched or not. B+ supply plays no part in this, even if you have heroic regulation. The fix is non-trivial and PCB isn't happy while dremel is singing. When measured in something like RMAA the above results to 1-2% IMD, but really it's much worse with one quarter of the waveform momentarily destroyed. RMAA IMD measurement takes several seconds so the waveform partially settles, which "improves" results. During 500ms it creates a sound actually quite similar to a fart. No joke!

Go ahead and try to collect all the info how to fix this one. Oh wait, nobody actually ever noticed it! For a long time I also thought it was just very low headroom of the gain reduction stage. Headroom is actually ample if only the design was complete and worked. But analag isn't exactly known for tested, complete and failure proof designs, he's more a spice-concept-guy with a strangely high reputation.

Now [silent:arts], your PCB design skills are visually remarkable. I would have still wished this design would have gone through rigorous testing. I realise no performance specs were ever published and that no one claimed this was some end-of-all vari-mu compressor. But the shortcomings were more serious than simple and innocent debugging errors.

And that still leaves us with the oscillation in the sidechain. And don't get me started on the PSU.
 
Kingston: This is interesting! I'm wondering why I don't get any audible farts when using it with a kick? With 100Hz sine wave and sudden increase of the level, no audible distortion. Maybe i'm doing something wrong? You would think that kind of a distortion would be audible, eh? I think I will have to check my PM670 again with an oscillator, maybe i'm able to find something...

What you're saying is, you're not going to tell what is the matter with the design although you know the answer?
 
That's interesting Kingston. I'd also really like to hear what you think is causing this distortion. I promise not to mass produce and sell the design!
 
telefunk said:
p2p? Why? there's nothing wrong with the excisting pcb's.

Why use a pcb for such a simple thing?   

What's wrong with the PCB's?  R spacing tight, only allows specific resistors.    If you like to use many specific types they hardly fit.  Forces use of DC for filaments, unless you want to properly hand wire separate AC filament, by which point you may as well go P2P for the whole thing.

I despise the PSU PCB layout.  A canyon between capacitor cliffs for all the small easily destructed parts one might need to change.  Should be the other way round,  big caps in the middle small guys on the outside where they are easily reached for service. 
 
telefunk said:
Kingston: This is interesting! I'm wondering why I don't get any audible farts when using it with a kick? With 100Hz sine wave and sudden increase of the level, no audible distortion. Maybe i'm doing something wrong?

Statistics:

audio by nature is almost randomly mixed waveforms between 20-20000hz. A quarter of the waveform is destroyed in and there is program dependency of sorts. Chances are you don't know what you should be listening and in fact most of the time this thing is near invisible due to the settling. Note, this problem is with the output stage only, nothing to do with the compressor program dependency. Maybe it cuts exactly the right amount of bass with your kick for example. At times it's like a ring modulator.

A great example just how bad distortion detectors our ears are due to their adaptive nature.

duantro said:
That's interesting Kingston. I'd also really like to hear what you think is causing this distortion.

There is a momentary DC charge at the output transformer on sudden RMS level changes. Creates this asymmetry. It shouldn't happen and lots of classics use this exact same output, but there it is and won't go away. Or maybe the classics have it too!  :eek: I'd be surprised, but it's not impossible.

I don't know the exact cause, but it needs a rearranging of the B+/plate configuration and a cap.

I said it yesterday already, I'm compiling a fully debugged revision and will post soon.
 
I ran electricity thru my PM670 and no Black smoke! Tubes glowed nicely. Next on to calibrating and tube testing. Meters are pegging to the right regardless of added parrallel resistance. I'll see after it's adjusted.
 
There is a momentary DC charge at the output transformer on sudden RMS level changes. Creates this asymmetry. It shouldn't happen and lots of classics use this exact same output, but there it is and won't go away. Or maybe the classics have it too!  :eek: I'd be surprised, but it's not impossible.

I don't know the exact cause, but it needs a rearranging of the B+/plate configuration and a cap.

I said it yesterday already, I'm compiling a fully debugged revision and will post soon.

I read this 7 times and can't figure out… can you please name the transformer with "charged DC on output"
 
Well, a very interesting description of this issue Kingston - thanks for taking the time to describe, in your own forthright, candid and even a little controversial style :) 

You simultaneously bring issues to light and make unique narrative around them, as well as those who brought them to us. :)

-------


I've always felt that my simple time constant implementation is probably not the best and letting the side down in terms of the grabbing/release characteristics. Resulting a little mangling of the waveshape.

A tendancy to be a little 'flabby' or with somewhat too much 'laissez-faire' attitude towards the grab and release.
Even on the quicker settings.

I have a feeling you probably may be right - I haven't noticed or checked this particular distortion you mention, so I think I'll take a look myself.

It's not easy to capture these sorts of dynamic behaviours so spending some more time is probably worthwhile now that I have a several year working unit.

------

That aside, I definately agree there are some distortions present in a typical working unit related to all manner of mismatches in tubes, transformers and such.

But my experience has not been that they are not that bad - I would characterise my units as having a mild but definite tube character and some graininess from the Edcors. Good for tracking and such.

I don't think I've come across a usage where I heard anything at all in the way of thump, however.
The actual GR performance is pretty transparent in sound but I do notice some unevenness is grab/release

And I agree these are not my choice for any kind of hi-fi mastering, although maybe on some mix stems would be good.
Particularly if looking for some .. warmth/distortion .. whatever one might call it.

But I certainly haven't so far noticed the obvious kinds of explosive flatulent events you refer to!

For sure, there is a strong measure of instability there, probably related to the amount of wiring and quantity of easily-interfered-with Edcor xsm transformers which seem to need quite optimal placement otherwise noise can take a hit.

And I have had lots of sessions retracing my steps and so on to eliminate issues when building.

-----

Another interesting alternative is baadcode's hybrid vari-gm with discretes and so on - the prr175 build might be an easier and more refined way to go. Both in terms of the sound and also the building effort.

Anyway, glad to have them and education that went with it. And that prepares me for the MkII hybrid pcb/pp approach I'm taking on my the next one.

The standard PM670 requires a will of iron to see through. Not at all for the faint of heart  :)

My own big-block dual unit cost about $1100us or so from memory, and  I would say around $600us for a single channel with good VU meter and Edcor signal traffos.

I do recommend doing a single unit(s) - quite a bit easier to get it right with only 1 in a box.
I think the stability and likelyhood to oscillate or noise out would be better too.
Not to mention that cash money part!

-----

However, mine work well, I use them alot - nice on mic'd acoustic guitar too - that is a hifi delicate application and it works very well for me. I consider them to be stable, trouble free and in the studio, am happy to reach for them :)
(Disclaimer - I'm not a studio professional ;D)

I wouldn't play a bass thru it unless I wanted some preamp-like brown crunch added - it sounds a bit like a Rickenbacker bass compared to a Fender bass. Flabbier and with some growl  :)

Happily, I happen to luv that sound on the bass, so I use it all the time for that!
For sure on all manner of eguitars. Even the piano especially for dirty rhodes and the like.

I have it in line with a G9 and dual Gpultec - all of it together still makes for a decently hifi sound if I want it.
And also very easy to drive harder for effects when desired. The relay bypasses on each unit mean I can switch them easily out of the chain for comparison.

The single channel of it I incorporated into a channel-strip-with-drive-stage really benefits from it - I feel it actually cleans up a crunched-up intrument sound so the overall thing sounds tighter and crisper.  Definately reminds me of tape effects - subtle but you know it's there and would miss it if  'twern't

cheers
 
Moby said:
I read this 7 times and can't figure out… can you please name the transformer with "charged DC on output"

Sorry for confusion, I think the correct term is DC 'offset'. Plate on side A is at a different potential than plate on side B (class A/B). I wrote "charge" because the offset is momentary, less than 500ms and goes away. Until you stop the music and let it "charge" again. The problem is somewhere in the tubes and their configuration because cathode bias is steady DC and the B+ at the centertap is always steady as well - both well regulated. Here some actual guru could come in and explain.

I wasn't that interested, because the obvious answer was adding a capacitor to stop this DC before the primary.
 
It's visible on any oscilloscope with fast enough screen update. And again, quite audible. Since that 2-4% IMD doesn't set off any alarms and people will simply shrug it off, I think I will slither back to the hole whence I came and leave everyone be.
 
Kingston said:
Moby said:
I read this 7 times and can't figure out… can you please name the transformer with "charged DC on output"

Sorry for confusion, I think the correct term is DC 'offset'. Plate on side A is at a different potential than plate on side B (class A/B). I wrote "charge" because the offset is momentary, less than 500ms and goes away. Until you stop the music and let it "charge" again. The problem is somewhere in the tubes and their configuration because cathode bias is steady DC and the B+ at the centertap is always steady as well - both well regulated. Here some actual guru could come in and explain.

I wasn't that interested, because the obvious answer was adding a capacitor to stop this DC before the primary.
If I understand the problem is in tubes current mismatch. Besides static match tubes must be dynamically matched for perfect behavior and minimum distortion. Also, Edcor transformers are really asymmetrical but you get what you pay. It works much better than it costs but far away from hi end like original Fairchild transformers. So, besides initial lie that tubes don't have to be matched everything else is transparent and true. Just my 0.2 c ;)
 
Moby said:
If I understand the problem is in tubes current mismatch.

It's not. I actually managed to match several quads to perfection. And I mean perfect in all points of the grid-voltage/plate-current curve between -2 to -12VDC at 0.5V steps. Overkill and you can probably imagine the amount of time this took for 60 tubes. I wanted to eliminate this avenue of error completely.

[edit]

I know this error is rather exotic and hard to understand (and to hear for everyone apparently), but quick changes of grid voltage do not generate any kinds of charges that last half a second and are not frequency dependent in any way. If this problem was plain tube mismatch, you would see the resulted cross-over distortion at the primary at all times. It does not die out in 500ms.
 
It's all very fascinating - measuring performance, establishing spec targets and so on.

I hope to understand more about the hidden levels this interesting limiter.

For sure I have no doubts that if rigorous performance testing was done on my one, it would come up more like a tube amp than a chip amp.

There has to be all kinds of raggedy things going on with all that wiring and wander-minded old tubes and those pretty cheap transformers.  My switches alone had mismatch from the resistors. But the tubes! Not even close. And if so, only for 3minutes.

Resulting  in tubes not opening up fast enough or at an even rate or having to work against a dc shift somewhere in the cathode circuit etc. And the discharge rates being lumpy or what ever.


I certainly saw lots of mismatched voltages and blips and blarps - when running pre-prepared test program material, not just sines.  ie. transients, pauses, different amplitudes and number of cycles, sweeps with transients etc.

So the questions of 'does it work' and 'how well does it work' is the main thing to decide on.
That can also depend on the accuracy of your measurement devices. How much is significant?

Makes you realise why commercial ones are priced the way they are. Takes a lot of hours to get right.

I would be very interested to play with a commercial vari-gm to hear and see for myself how good they can be when well maintained. Not likely!

Anyway, Kingston - for my part I would be happy to hear more about what you have found in the units performance when looking at the more detailed levels. 

I haven't optimised at all - no selecting of parts or tubes, really. Just the most rudimentary, neanderthal ham fisted checks of cathode voltages and grid voltages to get them in the ballpark. But I'm having a play through it today.

Fortunately I can test it all in place and see what I see and or hear.


I have a MkII to do and maybe I can avoid this issue there  :) 

Thanks

 
Back
Top