the Poor Man 660 support thread

[melville]

Cheers for the search tip.

Dominic

 

[radiance]

Light at the end indeed!! Tube matching finished and now my unit sounds wonderful.
I only had 5 6bc8's which were not all over the place, but somehow there was no match possible, no cross match neither  due to the first halves all being higher than the second.
However, the 6N5P were very consisted. I only had 8 of them but with cross matching them I had two decent pairs.
It sounds soo much better than before. Read Kinstons post below. It's exactly like he describes.
I still have to deal with some meter and knob interaction weirdness. Might have to do with transformer placement (to close to the meter) or maybe with the 5687's being weird them self.

With only 1 5687 inserted I have:
normal 6,5V heater voltage
No knob interaction with meter
Less distortion when time constant pot is turned CCW. When turned CW no distortion

Enter the second 5687 and I can't get my heater voltage above 6V, and knobs interact with the meter. This is not audible BTW.
I also have more distortion when time constant pot is turned CCW. When turned CW no distortion
This heater issue is strange as one 660 board heater is powered with the 6A secondaries from the EU groupbuy toroid, and the other is powered with another 8A toroid. Enough juice if you ask me....,

there's finally a light at the end of the tunnel.

I built lolo's grid-voltage/plate-current matcher. I went through a batch of 20 6BC8/6BC6, various brands. Then I went through a 12-batch of russian 6N5P, roughly an equivalent.

In a "pm670 environment", one never needs to match outside the grid voltage range of -1.5V to -8V. The curve will always fall there (as about 3dB to 25dB on the VU meter).

Anyway, out of the batch of 20 6BC8/6BC6, there was only a one "equal triodes" match. And that would not be a problem, but the rest of them were all over the place, and in the batch I could not even cross-match to get two equal triode channels of four tubes total. ie. cross matching triode sides to create a roughly equal current draw.

And I had actually found the best match with my own crude matching set up. It sounds like crap, with odd distortion, and no headroom, and unequal stereo image. That's what I've always been complaining about.

Enter 6N5P. Out of a batch of 12 there were four perfect matches, all values pretty much equal. With cross matching I can actually use most of the rest of the tubes in the future as well! Incredible tolerances! About a third of them have an odd unstable heater (no 6BC8 has this), and some of them can be heard doing a "shuffling paper near the mic" type of distortion. I'll test if these are simply duds, or if they just need a burn-in.

Headroom is finally there, now in the range of the rest of the studio! And the "roof" of the headroom is very very soft saturation, just wonderful! I've never heard anything like it. This will find a LOT of use. 6N5P is simply a great tube.



The unfortunate Edcor lack of quality can stil be heard, but that's an easier (if a little costly) fix. Lundahl to the rescue.



Now that the unit is in perfect working order, I found a new source of distortion, very much measurable: the VU meters. Is there some way to buffer up - or somehow isolate - the VU meters so that they don't affect the sound? They do odd things to midrange frequencies, and show up as added THD, in a slightly different shape than unbuffered VU meter on 600-ohm balanced outputs.

I guess an easy fix would be to put in a switch with a choice of VU meter and a dummy resistor equal to the VU load (about 50ohm for me).




[edit] while I was writing this it turned out russian the tubes just needed a burn in. No more "paper shuffling" distortion.

 

[Kingston]

Less distortion when time constant pot is turned CCW. When turned CW no distortion

this is normal. the stock time constant network is a bit odd.

Enter the second 5687 and I can't get my heater voltage above 6V, and knobs interact with the meter. This is not audible BTW.
I also have more distortion when time constant pot is turned CCW. When turned CW no distortion
This heater issue is strange as one 660 board heater is powered with the 6A secondaries from the EU groupbuy toroid, and the other is powered with another 8A toroid. Enough juice if you ask me....,

I forget, are you using the regulator PCB or are you just dropping heater voltage with a high watt resistor?

 

[radiance]

I forget, are you using the regulator PCB or are you just dropping heater voltage with a high watt resistor?

regulated....

 

[Kingston]

Your issue might be the regulator overheating. Needs a pretty substantial heatsink for stable use. Then check how much those 5W resistors around the regulators are dropping. That's a great indication of many things.

Also there has been issues with counterfeit LM338 (LM317 marked as LM338) in Europe.

 

[radiance]

I measured the  temperature of the heat sinks very close to where the regulators touch them and it was 60C, so I guess the regulators are 70C. That's within spec for both the 338 and 317. These things can operate up till a friggin cooking 125C  
Also, with two 5687's inserted I measure 2,6A current draw for 1 PM660 board (@ 6V). That's almost twice of what the 317 is capable of.
What would actually happen when a 317 shuts down or has to regulate to much current?

Then check how much those 5W resistors around the regulators are dropping.

Can you elaborate?

 

[Moby]

I measured the  temperature of the heat sinks very close to where the regulators tough them and it was 60C, so I guess the regulators are 70C. That's within spec for both the 338 and 317

Yes, it can be true if the thermal connection is great. If You have poor thermal connection between regulator and heatsing it's probably much, much more than 70C. 

 

[Kingston]

What would actually happen when a 317 shuts down or has to regulate to much current?

When it "shuts down" from overheat or over-current it still lets some 1V through, trying to protect itself.

Then check how much those 5W resistors around the regulators are dropping.

Can you elaborate?

Just meant that check all the voltage drops around the regulator. Also check how much the regulator is dropping, and exactly what voltage goes in when it can't deliver those said 6.3V for the two 5687.

 

[radiance]

Yes, it can be true if the thermal connection is great. If You have poor thermal connection between regulator and heatsing it's probably much, much more than 70C. 

They're isolated from the heatsink with a silicone (is it silicone?) patch. Dunno if this gives a proper thermal connection.
Here some pics.

As you can see, the heatsink is attached to the chassis by means of a solid piece of aluminium

When it "shuts down" from overheat or over-current it still lets some 1V through, trying to protect itself.

Well, that's obviously not the case...

Just meant that check all the voltage drops around the regulator. Also check how much the regulator is dropping, and exactly what voltage goes in when it can't deliver those said 6.3V for the two 5687.

I did some measurements.
After the rectefier I've 9,9V DC > after R1 9,3V DC > after regulator 6,7V DC > after R4 6V DC.... & 0,750V AC (1000uF caps not replaced with bigger ones....)

 

[Kingston]

I did some measurements.
After the rectefier I've 9,9V DC > after R1 9,3V DC > after regulator 6,7V DC > after R4 6V DC.... & 0,750V AC (1000uF caps not replaced with bigger ones....)

That's perfectly normal. You can bypass R4 for example, and just set regulator directly 6,3V. Or make R4 smaller so you will land at about 6.3 ( this is what I did for example). Those 1000uF absolutely need to be bigger ones for this amount of current. 0,750V AC is normal with the small ones. All of them need to be 4700uF but preferably even 10000uF. Expensive, but there's no other way around this one if you want clean DC with this amount of current. That PCB is basically a bit of a flop, since it was specifically designed for high current heater regulation. The big caps dangle around in an ugly (and unsafe) way from this small board.

 

[radiance]

The big caps dangle around in an ugly (and unsafe) way from this small board.

At RS, I found some Panansonic caps from 4700uF  16V measuring 12,5 mm diameter and 40 mm high. These should fit the pcb but like you said, are expensive. Is perfect DC really necessary?

Also, I suppose R4 it's purpose is current limiting..but what for? What's the idea behind R4 and is it wise to bypass it ?

 

[Kingston]

Is perfect DC really necessary?

Also, I suppose R4 it's purpose is current limiting..but what for? What's the idea behind R4 and is it wise to bypass it ?

of course <1mVAC ripple on the heaters isn't necessary and you should really test first if that 0,750VAC leaks to audio at all. But we already have a regulator so 0,750VAC from this board would be a complete waste of it.

R4 prevents a super-rare case of oscillation with the regulator, and also removes some of the white noise from the regulator. It's actually better to bypass R1. It's there for a marginal extra smoothing only. was never needed in the first place since we have the regulator already, especially this low resistance.

 

[radiance]

It's actually better to bypass R1. It's there for a marginal extra smoothing only. was never needed in the first place since we have the regulator already, especially this low resistance.

Thanks! I'll try bypassing R1....

 

[opacheco]

[quote author="[silent:arts]"]

[quote author="rrs"]Also looks like I need to solder some resistors to these suckers. Is anyone able to help regarding values and working out where the go??

look at page one in this thread, there is a link regarding the switches and resistor values.[/quote]

Log Pot to Switch Excell Sheet

Thanks  :thumb:

I have the 12 pos 2 deck for threshold and gain Here. So just wondering how to approach that (with the 2 deck) ? Also what wattage do these resistors need to be?
[/quote]

rrs, thge link is dead now!!...do you have a active link???

Thanks

 

[[silent:arts]]

always look in the first post from this thread
Log Pot to Switch Excell Sheet

 

[Gus]

No one sees the issue with the heatsinks?

First the plain Al heatsink could be too small,  AND the adding of the two small heatsinks is probably worse than not using the two small ones you have added two more interfaces to conduct heat not a good thing.

Note the heatsink will not be as hot as the IC chip so the heatsink being 60C does not tell you the IC die temp.  The thermal interface (you have two for each reg IC)can fool you because the interface might not be transferring the heat away fast enough casing the heatsink to not get as hot as it should.  You also need to be careful when you tighten the reg you can bend the tab and reduce the transfer of heat.  

http://www.aztronic.cn/docs/72674/72674.pdf  Torque spec TO220

http://www.littelfuse.com/data/en/Application_Notes/AN1004.pdf   page 4

I am not sure what the circuit is, is it 317 based?

If 317 it is only good for 1.5 amps with a good size heatsink and not too much power being "dropped" across the 317

http://www.national.com/ds/LM/LM117.pdf

http://www.national.com/ds/LM/LM138.pdf

 

[syn]

Sand the joining surface of the small heat sink with fine sand paper, sand the joining surface of the large heat sink with fine sand paper, apply heat transfer compound to either, join tight, it should improve heat transfer, dramatically, hopefully. The glossier the surface of the heat sink is, the better the heat transfer is. Hope it helps.
Those pads that you use do not transfer heat very well, also, I would use two of each, not two small and one large heat sink.
EDIT: or loose the small ones and go with large ones only. The sand thing is still worth trying. I am aware that this suggestion complicates electrical insulation, but it is not impossible to do.

 

[Gus]

Sanding the surfaces can make it worse.  The pads can work OK and are less messy.

 

[radiance]

Thanks form the comments!
The regulators are LM338's. All heat conducting surfaces have been sanded/polished.
I'll get rid of the two small heat sinks, see if that'll make a difference. This makes sense...

However, I'm not convinced that there's a heat problem as far as the regulators are concerned.
First of all, when regulators suffer from thermal shut-down they will not deliver 6V @ 2,6 A right?
If they did suffer they would be fine during start-up, maybe several minutes before they shut down.
Or is there a stage where the regulators are not shutting down but are unable / barely able to deliver the current due to ALMOST overheating?
I've had the poorman on for about 45 minutes with the case closed and the big aluminium heat-sink did not get hotter than 60C.
I know (now) that the regulators get much hotter, but HOW much? Twice as hot maybe? That would be 120C, uncomfortably within spec  

 

[radiance]

Found a nice, bigger and one with way more surface heat sink.
Will use that one just for the sake of things will run cooler.
And, somehow I DO like metal work....