PMsix61 Limiter DIY

[DaveP]

It was 4mA/Tube, 16/side, 32mA total.
best
DaveP

 

[alexc]

No, I got that - I was wondering about the gain.

Cheers

 

[DaveP]

Alex,
I have not fired up the signal amp yet.  I nearly finished it then realised that I did not know enough to set it up properly.
That's when I started to make other vari-mu's first. I have the very expensive Sowter TX.
However, I have no reason to think that the output would be any different from the stock Fairchild which is +16dB for the 660 and +8dB for the 670 which has half the tubes.
best
DaveP

 

[alexc]

My understanding was the 660 has 2 6386 per phase same as one channel of the 670.
I checked the schematics and it does seem to be the case 4 tubes per signal amp in both cases.

I did know there is a difference in overall gain but I thought it related to the bias differences.

Still, I don;t have any experience with the real deal so I could be wrong.

Thanks

 

[DaveP]

The number of 6386 tubes is the same for 660 and 670, but there are two/phase/channel in the 670 and 4/phase in the single channel 660, double the tubes so double the gm and twice the gain.
best
DaveP

 

[alexc]

I must be confused and if so I do apologise - but there are a total of 8x 6386 tubes in a 670 and 4x 6386 in a  660.

Maybe you are referring to the M-S type matrix operation thing in the 670?

 

[DaveP]

Sorry, you're right, the 670 schematic only shows two/phase for clarity but puts an asterisk to show they are doubles.
best
DaveP

 

[MatthisD]

But I'm guessing I have that wrong - you indicate each phase is 2K. Which would make it worse.
I was going to try to load the secondary higher, but that seems to be causing more probs in other areas right now.

If the impedance ratio of the transformer is 8K:600 the voltage ratio is 3.65:1, the primary is centre tapped giving 1.825+1.825:1 which is an impedance ratio of 2K+2K:600.
  A 10k pot across the output could be used to find the best loading (using wiper and one end, the other end not connected).

Edit:
I had a look at the 660 schematic, the secondary loading is 1.86K in parallel with the load on the line out connection.

Sowter list there signal output as being 9+9:1+1, compared to 5.5+5.5:1+1 but in a single channel the Sowter can be wired 9CT:1 (secondaries in series) which is just a dB or 2 difference and overall gain should equal the 670.

 

[vari-mu]

Sowter list there signal output as being 9+9:1+1, compared to 5.5+5.5:1+1 but in a single channel the Sowter can be wired 9CT:1 (secondaries in series) which is just a dB or 2 difference and overall gain should equal the 670.

Original infos output trsnsformer TRIAD HS-52 :
20K/600 ,  5.7:1

 

[alexc]

If the impedance ratio of the transformer is 8K:600 the voltage ratio is 3.65:1, the primary is centre tapped giving 1.825+1.825:1 which is an impedance ratio of 2K+2K:600.

That's a way to look at it. I have been following some pp tube guides and drawing up some load lines which go along the lines of 1/2 the plate-plate impedance per phase for class A, which occurs when both phases of the primary are conducting  and 1/4 for class B, which occurs when one of the primary phases is not conducting at all.

So far, I have been running class A. But anyhow, another item to file upstairs as I go on learning.

I'll give the potentiometer measurement a go out of interest. By 'optimum' you mean max power transfer ?  And that would be at the 'idle' or min GR condition, when the plate resistances are lowest.
So when in deep GR and the plate resistances are highest it is least optimum, I guess.

I'm also just wiring up the signal amp of a couple of rca type limiters using these traffos but driven with more conventional cathode bias 6V6 pp pair. I'm not expecting any matching issues here with 600R term.

Thnx

 

[alexc]

Interesting that HS52 is - with 20K primary it has a 12.5mA (balanced) dc spec.
Whereas the 670 seems to run an idle of around 24mA or so.

So the basic gain structure of a 670 would be  of 1:9 (+19dB) at the input, then whatever the 6386s do and 5.7:1 (-15dB) giving a total max of +8dB as specified as the gain  ?

6386s into the HS52 being +4dB  or x1.6 or so  ?

 

[alexc]

I've been describing the noise testing that I'm doing in my dual rca build in my other thread.
It uses the same psu traffo, output traffo and rack with a second channel even closer to the psu traffo.

Early indications are good - as long as 600R secondary termination and 8K primary termination are valid, no-load-PSU noise levels are tiny. I'll be testing it as I go to try to avoid the problem I have in this build

So far, so good and making me feel better about the output traffo. We shall see!

 

[MatthisD]

Sowter list there signal output as being 9+9:1+1, compared to 5.5+5.5:1+1 but in a single channel the Sowter can be wired 9CT:1 (secondaries in series) which is just a dB or 2 difference and overall gain should equal the 670.

Having looked at the schematic to follow the connections from the signal amp's outputs to the switch network for channel functioning, the secondaries are in series for independent channel functioning. That changes the ratio I mentioned above to 5.5+5.5:2 which is 2.75+2.75:1.

To simplify it, I'll ignore the 600ohm load that would be connected to the output. Theres 980R (round up to 1K) placed across each secondary half. Each of those halves are a turns ratio of 1:11 to the total primary. 11x11x1K=121K (round to 120K), two loads of 120K in parallel=60K. 60K is the load to the total primary. Ignore the DC voltage and separate each primary half into two windings. Then each of those are 15K windings, dictated by the turns ratio and secondary loads.

Connect a 600ohm line load to the signal amp output which is across both secondaries in series and recalculate the load; 5.5x5.5x600R=18K to the total primary, 18K in parallel with 60K from above = 14K (with some rounding). The two primary windings are now 3.5K each.

About the HS-52; 5.5:1, 5.6:1 or 5.7:1 is much the same 20K:600ohm if using round numbers.
  The terminal numbers on the transformer correspond to the windings on the schematic to indicate 5K+5K:150ohms+150ohms rather than any other configuration of taps or impedance connections unless that is a coincidence.
  12.5mA DC current rating; a wire gauge chart matches 17.5mA to 39AWG which is what the primary is wound with according to previous threads. Triad used a safety margin perhaps, current could surely exceed 17.5mA by a couple of mA without failure. Dave posted 16mA per side.

Transformer gain and loss with input primaries in series and output secondaries in series;
Input =1:4.5+4.5 output=2.75+2.75:1

In this case the Sowter output, listed as 9+9:1+1 = 4.5+4.5:1 which would yield a few dB more loss overall and with less load to the primary.

By optimum loading using the pot I don't mean anything in particular just a way to quickly vary the load to experiment in case of over or under-loading.

 

[alexc]

Thanks MatthisD

That's a really informative analysis and one which I will go thru slowly with the schematic in front 

 

[alexc]

I finally found the source of most of the ground noise - the heater regulator module.
Damned if the thing isn't making one hell of a hash of things!

So reworking all the heater circuit and that should help.
Without the reg and a dc with the current huge ripple my SPL noise floor is degraded by 5dB for 12dB gain vs the previous 21dB.

OK - so on the path back now

 

[DaveP]

Alex,
A simple schottky diode pair with a centre tapped TX followed by 2 x 47,000uF 16V Caps with an Al clad res between them works fine, fit and forget.
I subscribe to the "keep it simple" mantra 
best
DaveP

 

[alexc]

Good advice. I'm going to ditch the regulator module.  ;D

I don't usually use them in these sort of circuits, but this time I thought I'd be more elaborate 

I don't have a CT winding and I already have a high power bridge in place for the control amp heater.
So I'll add a new circuit to it with 20KuF -> dropping resistor ->  another 20KuF. 40KuF

I have 2.4A in this circuit so that should get me < 25mV of ripple.

Once that is tamed, I can continue to check the grounds for noise till I get the signal amp nice and clean.

Then, bump up the current again  ;D and stay with the 8K:600. (haven't ordered a replacement yet)
It seems to work fine in terms of handling the min GR to max GR transitions.

Then I'll just have the control amp instability and then some suspected signal-control cross talk issues to resolve 

 

[alexc]

One ordeal of a day - I reworked the heater supply to a reasonable 50mV of ripple with 20KuF+0.75R+40KuF.
Could probably get it down a bit more, but that's OK.

But still no cigar! Noise stubbornly remaining at -59dBu overall.  Well several hours of stripping back the circuit till I isolate it down to my UTC input traffo 500:15K. I'll be damned but it's incredibly sensitive to orientation
I had it writing to front and changed it to writing at right. Difference was 17dB of mains harmonics all the way up to 2KHz

So now I have the same circuit as I always had  but I have an SPL noise floor of -74.0dBu from a loopback of -78.4dBu calibrated to voltage and a 'power off' of -78.2dBu. That's with the signal amp with a net gain of +8dB.

So 4.2dB of added noise for 8dB gain. Which is OK.

Grounding the primary of the input traffo has no effect. Grounding grids gives -75.2dB
Now I still have the wiring to re-button down so I think I can probably gain another dB or so.

Interestingly, those figures are with the input connected from the Motu, no signal applied and the 'Level' T-pad attenuator at minimum level (max atten). If I open it up all the way (min atten) then the noise floor drops to -76dBu and a typical setting around the middle gives -76.2dBu.

That's 50Hz at -96dBu, 100Hz at -96dBu and 150Hz at -84dBu. The harmonics disappear at about 250Hz.

Anyway, that's that. Tomorrow to rebutton everything up, checking as I go then amping the current up 'gain.

God what an ordeal! But all's well now  I don't have to spend up big for better traffos!

 

[MartyMart]

So great that you have that problem solved Alex

I find it quite amazing how the simple orientation of I/O/PWR transformers can make such a
HUGE difference !

Humans win

Marty.

 

[DaveP]

Yes, full marks for perseverance Alex, its the only way ;D
best
DaveP