Gates Radio SA-39 limiter restoration and modification

[cuibono]

Hi, I'm restoring a Gates SA-39, and had a couple of questions.

First off, most of my B+ voltages are about half what they are supposed to be.  This seems to be because of a large, 1500ohm 10W (approximately) resistor in series with the power supply inductor L1 and the 6L6G tube (pins 3 and 4).  This resistor isn't anywhere on the schematics I've seen, but it looks original.  It looks like its been getting hot, but isn't toasted.  Because all the tubes aren't getting a full supply voltage, the front panel meter can't reach 0dB.  So what I'm wondering is, can I put a smaller resistor in, like 500ohms? 

Similarly, there is no R47 across the meter, and I don't see signs that there ever was one.  This model seems to have been designed a little differently than others.  Has anyone seen this before?

Thanks!

 

[emrr]

You have other problems.  Yours is an earlier schematic, and those parts are correct. 

Did you adjust the regulated supply properly?  Have you measured tube currents yet? 

 

[cuibono]

Ah you beat me to it!  I was about to say I traced the circuit wrong.  It turns out R44 is only measuring 7k, so it need replacing.  Duh.

I've been matching tubes - how close should they be?  5% of each other?

 

[emrr]

Match them with the current jacks on the sa39.  I'm on a phone and have no idea what r44 is supposed to be.  There are some large power R's in the early versions that always measure wrong in circuit, and fine out.

 

[emrr]

The terminal board layout differs on the early versions as well.  The first version SA-39 is extremely different in circuit. 

 

[cuibono]

Thanks for the help - should I match the the current within 5%?

 

[emrr]

Current on the 1612's should match as close as you can get them with no gain reduction, AND with some amount of GR.  I look at 0, -5, and -10 dB GR.    Current on the 6V6's is the only other that matters very much, and that's easy; buy new matched tubes.  The Tube Amp Doctor matched sets are good and very reasonable.    If you are matching yourself, get as close as you can.  The closer you get, the better and cleaner your low end will be.  Typical 5% would be 38 and 40 mA.  You can buy tubes that are dead on 40/40 as new pairs. 

 

[cuibono]

Very helpful, thanks so much!

 

[cuibono]

So I'm almost there.  The signal looks good coming out, and I see obvious compression when pushed past a certain point. 

It looks like the meter isn't really doing what it is supposed to though.  The meter sits at -3dB after warming up, and when there is obvious compression on the output (as seen on a scope), the meter reads at most 2dB of compression.  More like 1dB.  My unit has no resistor across the meter (R47).  The schematic says its value must be determined by tube currents, but gives no formula.

Here are my tube currents:
V1: 2450uA
V2: 2430uA
V3: 940uA
V4: 870uA
V5: 40.55mA
V6: 40.30mA

Tubes 3 and 4 are a little short of the 1100uA specified, so I could try to switching them out.  The regulated B+ sits around 276V, +/- 2V.  I have adjusted R3 for maximum current through the tubes.

Any ideas?

 

[emrr]

Eliminating assumptions; you've maxed out the 0 adjust pot to get that reading?

So we don't have enough current for the meter to hit 0.  This is pretty common, I've seen it frequently. 

If you add a resistor across the meter, you bypass some current by the meter in parallel, right?  You get an even lower max reading.  Not the problem. 

So is anything in V1/2 path reading much higher resistance than spec?  R2-R7.  Red herring; you have official current out of V1/2.

The simple cheat of course is to lower the value of R2 for appropriate range on the 0 adjust.

 

[77 Cathode Strip]

Wow, seems I got to this thread just in time.  I have an SA-39b but with an SA-38 meter and my meter also reads low with the R3 pot fully turned.  I have a couple of questions but first some stats.  I have two NOS Ken-Rad 6L7s subbing for the 1612s.  Plate load resistors read R5: 84k and R7: 87k.  Plate voltage is V1: 174 and V2: 162.  Cathode R2 is 509. Cathode current is V1: 2.32ma and V2: 2.27ma no signal.  R4: 66.5k, R6: 48 ohms.  Regulated rail set to 275VDC.  Screen supply is 50.1.  New 6X5, new 6H6, new Sovtek 5881 for the 6L6 (also have other 6L6s).  New NOS Westinghouse 6V6s.  Fresh filter caps in the power supply.  The neon bulb seems to be functioning normally.  Mine is also missing the R47 shunt resistor, but too much current at the meter is not my problem right now.

I have a couple of questions.
The SA-38 meter says "0dB = 1ma into 600 ohms".  The schematic for the SA-39 states the meter is 1 mil dc movement.  Same?

The filaments of the 6L7s have the DC supply and the voltage is low at 4.5 volts.  Should it be closer to 6 volts with the half-wave rectifier circuit or is it supposed to be that low? 

Thank you.

 

[emrr]

filament is half wave rectification, there's still an AC component you can measure. 

So someone hacked a 38 meter into a 39?  Did they also add the meter switching?  There is no difference in implementation of the gain reduction setting between the 38 and the 39.  The 38 meter has AC signal inputs and DC current inputs, I assume DC bypassing a rectifier for the VU portion.  I don't know if it's a 1mA meter, or a real 200 microamp VU with shunt to handle 1mA current.  Likely the latter. 

It has been rare in my experience to see one of these with the meter shunt resistor installed.  Maybe one out of a dozen.  I'm not in front of a schematic at the moment to compare stats, but they should be mostly self explanatory and easy to compare. 

If the 38 meter has an internal shunt that's fallen in value, it might give this problem.  But, that doesn't apply to a 39 meter, and it's an 'if'. 

The meter 0 point is largely of convenience, and GR readings are pretty loosely interpreted in this sort of unit.  Lowering the value of R2 has not ever hurt anything in any unit I've seen.  The other option would be to consider that some/all related R's may be changing to higher values in operation, and just rebuild that section with metal film.  Consider the 0 rheostat may not actually be going to a zero value, and thus not touching 509r as you'd assume. 

 

[77 Cathode Strip]

filament is half wave rectification, there's still an AC component you can measure. 

 

Oh, yeah.  3.39 AC at one pin and the 4-whatever DC at the other.

So someone hacked a 38 meter into a 39?  Did they also add the meter switching?  There is no difference in implementation of the gain reduction setting between the 38 and the 39.  The 38 meter has AC signal inputs and DC current inputs, I assume DC bypassing a rectifier for the VU portion.  I don't know if it's a 1mA meter, or a real 200 microamp VU with shunt to handle 1mA current.  Likely the latter. 

That last makes sense.  The 38 meter does have internal resistors.  Lifting them from the circuit was easy but the result was negligable.  Still sits at -3 at idle.  Since the rest of the circuit is nominally correct, it must be a difference in the meter movement itself.  Bummer.

The meter 0 point is largely of convenience, and GR readings are pretty loosely interpreted in this sort of unit.  Lowering the value of R2 has not ever hurt anything in any unit I've seen.  The other option would be to consider that some/all related R's may be changing to higher values in operation, and just rebuild that section with metal film.  Consider the 0 rheostat may not actually be going to a zero value, and thus not touching 509r as you'd assume.

  Simply grounding the high side of the pot takes it out completely.  To get the 38 meter to get near zero I would have to ground the cathodes and increase the plate load resistors and then the idle current and gain structure of the circuit would be much higher than designed.  I think I'll try to find another meter.  I could go ahread and install the attack, threshold and release mods as I search for a suitable sized and configured meter.  The 38 meter works fine although the paint of the face is a little flakey so when I am done maybe someone else can use it.

 

[emrr]

I don't know if it's a 1mA meter, or a real 200 microamp VU with shunt to handle 1mA current.  Likely the latter. 

That last makes sense.  The 38 meter does have internal resistors.  Lifting them from the circuit was easy but the result was negligable.  Still sits at -3 at idle.  Since the rest of the circuit is nominally correct, it must be a difference in the meter movement itself.  Bummer.

Simply grounding the high side of the pot takes it out completely.  To get the 38 meter to get near zero I would have to ground the cathodes and increase the plate load resistors and then the idle current and gain structure of the circuit would be much higher than designed.  I think I'll try to find another meter.  I could go ahread and install the attack, threshold and release mods as I search for a suitable sized and configured meter.  The 38 meter works fine although the paint of the face is a little flakey so when I am done maybe someone else can use it.

#1: an SA-38 meter is an unobtainium part worth having repaired properly, if damaged or out of spec.  It can't be had otherwise, and it's probably a bit criminal to dispose of one in any condition. 

#2: gotta ask; have you adjusted the meter trim on the face of the meter, and made sure it is aligned properly with the internal adjustment parts?

#3: Have you bypassed R2 and adjusted R3 to see where zero lands?  Measured the resulting cathode value?  That's the first fact needed.  Then, how does it sound and perform?  I think you are vastly over thinking what needs to be done here. 

#4:  Test the meter out of circuit.  Create a known 1mA current with a rheostat and DC power source, as measured on a multi-meter or other known good meter, then put the SA-38 gain reduction contacts in that circuit and see how it's calibration compares. 

#5:  you didn't really say where the internal meter resistors are exactly, do you know?  Are they in the DC or AC portion of the meter?  Connected how?  Is VU mode included in this mod?  If so how does it work in VU mode?  If VU mode included, are you sure it's hooked up to the correct terminals?  There should be a spare set.

Lastly, again, lots of 38's and 39's don't trim to zero with the stock arrangement anymore, and I've seen all of them brought to zero successfully with cathode R modification.  No other operational or sonic problems noted.  These things don't sit around at idle current anyway, they are always modulated by the DC bias of the side chain.  You can't really look at them as you would a purely cathode biased system.  There are early RCA limiters with similar input stages, having zero adjust that takes the cathodes all the way to ground. 

 

[77 Cathode Strip]

I'm sorry, I though I made it clear.  I am unable to zero the SA38 meter even with the cathodes grounded.  This is unacceptable.

 

[emrr]

Sorry, I wasn't clear on that.  It sounded speculative.  I'd confirm meter cal out of circuit at that point, or sub in another 1mA meter and see if it performs better. 

 

[cuibono]

The simple cheat of course is to lower the value of R2 for appropriate range on the 0 adjust.

All very accurate.  I will adjust R2 - but still the problem persists with the meter not showing compression reduction.

In fact, I suspect I'm just clipping the output, not actually compressing.  When I drive the signal about 6dB past the point where I can first see the waveform change shape, it rapidly becomes a square wave.  This makes me suspect that the signal isn't being compressed, just overdriven.  I replaced the 6h6 tube, because it had gone bad.  How do I verify the signal is being compressed?

Thanks again!

 

[emrr]

Take the 6H6 out, if it acts the same it's not compressing. 

Verify 6H6 is getting proper voltages, that signal is reaching it, and that you measure DC CV coming reaching the CT of the input trans sec. 

 

[77 Cathode Strip]

Sorry, I wasn't clear on that.  It sounded speculative.  I'd confirm meter cal out of circuit at that point, or sub in another 1mA meter and see if it performs better.

I'll have to pick up another suitable VU meter.  But as I was looking to see if I had one I found a nice old Simpson model 27 DC ma meter with a 1ma scale.  Subbed it in and by varying R3 I was able to get a swing of .6ma to .9ma.

I'll keep an eye out for another VU and I will proceed to making the mods.  I have searched a lot online and I have not found an SA-38 schematic to compare the meer circuit.  The shunt is internal to the meter, I will have to open it again to take a picture and try to measure the resistors. Thanks for your help!

 

[Donttry]

http://bit.ly/1arTLNl

Recent eBay listing of a Gates SA-39B with a replaced (output?) transformer.
Anyone know whether this is a big change in sonic character?