RCA OP-6 Analysis

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
V2 test.

B+ 248V, Vp=65V, Vg2 = 29V and Vk = 1.3V.

The gain with no load is 60 and with 1M load it's 46.  The gain is low because there is no bypass cap.  This is reflected in the rp which is 863.5k and the gm which is a very low 0.1973 making the mu 170.

The gain for V2 before the feedback loop is connected is therefore 46 or  +33.2dB

That's all for tonight

best
DaveP
 
DaveP said:
OK, jazbo8, here are your results:- <snip>
This gave rp as 368.4k and a gm of 1.224 (book says 1.185) I have a good tube I guess, others may vary.  mu would then be 450.
Thanks for taking the time to do this, while the gm is pretty close, rp is quite far from the "about 1M" given in the datasheet, which of course affects the mu. I will see if I can get my hands on some 6J7's for some tests... Since a single sample does not really tell us much.  :-\ In the meantime, I will follow this fascinating thread.
 
> rp is quite far from the "about 1M" given in the datasheet, which of course affects the mu.

In resistance-coupled pentode, Rp and plate Mu hardly matter.

Your op-point suggests 2mA. Say you take 100V drop across plate resistor. Plate resistor must be 100V/2mA or 50K. There will be additional loading from next-stage. Whether internal plate resistance is 380K (Mu= 470) or "1Meg" (Mu= 1200) does not really matter, if stage gain is 61 at best. 380K or 1Meg means 0.42dB difference.

You *may* need better than 0.42dB gain consistency. (Not in the studio, but cross-country telephone lines might have 50 or 100 repeaters in cascade.) That's why Mr Black gave us Negative Feedback.

Still in all, 380K smells low. While 6J7 was not sold as an RF tube, similar tubes were. Radio tuned circuit gain and selectivity is limited by Q. Q is limited by coil resistance and tube internal impedance. Coil losses are reduced by working the tuned circuit at high impedance, until the tube bites Q as bad as the coil does. RF/IF pentodes reliably gave Rp>1Meg, for Q>100 with coil total losses near 100 Ohms. The 6J7 does not have to meet that spec, but it is odd for it to be that much lower. But at audio frequencies it is a hard thing to measure.
 
Yes, thanks for the reminder on rp. It should not matter much, but the 300k-500k calculated values all seem a bit low, so I wonder if it is just due to the samples that Dave have on hand, or that's the way it should be... 
 
calculated values all seem a bit low

I guess that the greater the influence of g2 on the performance, the more the gain gets closer to that of a triode, which has much lower rp.

There is also the resistance of my scope probe?  I'm not sure how much that affects the result?  If the scope probe is 1M and I load the tube with 1M then we are down to 500k load.

The tubes I'm using are NOS RCA made in Harrison NJ with 230 in small numbers printed on the side.

best
DaveP
 
Now we come to the feedback in V2 and V3.

V2 gain is 46 and I estimated V3 gain as 80 so multiplied together they have a combined open loop gain of 3680 or
71.3dB.

When the later version 680k feedback resistor is in place, gain should fall to 3680/(1+(3680 x 2.7/682.7)) = 236.6 or 47.5dB.

Making 24dB of feedback.

To be thorough I decided to make V3 into an exact 80 gain tube by changing the resistors, having done that I connected the feedback resistor and the gain came out as 233 which is 47.3dB, so that checks out OK.

Moving on to the transformers:

The IPT has a dual input of 30 ohms and 250 ohms to a 50k secondary.  Allowing for transformer losses this works out as +22dB for the 250 input and +31dB for the 30 input.

The OPT is 48k primary to 600 ohms and with transformer losses it works out as -20dB.

The first snag!

EMRR measured the gain from V2 input to the output as 38.5dB, but from my figures it should be 47.3-20 = 27.3dB

Anyway, pressing on, we can start to estimate the overall gain.

V1 gain is 19.75 with its feedback loop in place which is +25.9dB
V2/V3 gain is + 47.3dB
IPT gain is +22 or +31dB
OPT loss is -20dB

So the overall gain is 75.2dB for the 250 ohm input and 84.2dB for the 30 ohm input.

But if we use EMRR's figure of 38.5dB for V2/V3/OPT with V1 and IPT figures we get:-

38.5 + 25.9 + 22dB = 86.4dB for the 250 ohm input and 95.4dB for the 30 ohm input, which is kind of what we were expecting.

So something is clearly amiss.

The gain of V2/V3 is fixed by the feedback loop so I wonder if the OPT has a different ratio.  If it was a more normal 15k to 600 then the loss would be more like 15dB instead of 20dB.  This would make my calculation +80dB for the 250 input and 89dB for the 30 ohm input.

Any ideas out there?

Best
DaveP
 
DaveP said:
But if we use EMRR's figure of 38.5dB for V2/V3/OPT with V1 and IPT figures we get:-

38.5 + 25.9 + 22dB = 86.4dB for the 250 ohm input and 95.4dB for the 30 ohm input, which is kind of what we were expecting.

So something is clearly amiss.

I measured the following, no search for mods or out of spec parts, no idea which version of parts values.
V2G to V2P: 24.5 dB
V3G to V3P:  34 dB
T2 : -20 dB
= 38.5 dB gain in output amp. 

There's also the output pad to account for, couple dB.

Remember 1st stage feedback changes, it's only max at one step from full off. 
 
> resistance of my scope probe?

Directly part of the measurement.

If the tube is really 613K, and you shunt with a 1Meg probe, the apparent impedance is 380K.

I don't know what rig you are using to measure.

Or what voltages you are reading. Most good 'scopes, the 10:1 probe will get you to 10Meg input. If signal levels are small, a cap-coupled TL072 with bootstrap bias ought to put measurement loading up over 100Meg.

Obviously ANY transformer (even a 445KC tank) has losses similar or larger than the tube.

But 680K seems plausible, 380K is just low enough to cause doubts.
____________________________________________________________

> When the later version 680k feedback resistor is in place

I see OT primary tied-back to V2 cathode? Isn't this a second NFB loop? However highly dependent on loading. (The "48K" is for design load, will be effectively larger no-load...) Something clever designed-in here.
____________________________________________________________

> more normal 15k to 600 then the loss would be more like 15dB instead of 20dB.

14dB (I had to check myself).

And if it were 15K, with RCA's budget, it would not need that monster choke.

And a happy condition for 250V and 15K load is more like 16mA than the 4mA-7mA we seem to have here. (Look at 6V6 Champ: 250-350V, 5K or 7K load, usually run at 35mA-40mA. Double 7K to 15K, expect half the current, 18-20mA, not 6mA.)

Yes, RCA could have run a heavy load with the skinny 6J7. There's higher vision, there's market needs (not too many tube-types), and there were the bowels of the Camden factory (I was in there, in the last days of such like this). Funny things happen to designs.
 
Doh, my mistake ::)

I forgot to account for the resistance of my scope probe which is 1M.

So jazzbo8's figures are now :-

rp=583.3k, gm=1.224, mu = 714

Sorry about that, we live and learn!

DaveP
 
Thanks for your input PRR.

I'm using a Tenma scope and sig gen, but from now on I will allow for the probe!

I was measuring output from the plate via the cap.

Yes, its a second feedback loop, partially by-passed for top end FR correction, that may be significant .

I think they used the choke at 6mA  to save battery life in a portable device.

I will investigate EMRR's figures.  They are taken within that feedback loop so I will have to calculate the open loop gain.

Best
DaveP
 
Thanks for the update, but I think with the 10x probe, the input impedance is 10MΩ. Anyway, since I still couldn't wrap my head around the relationship between rp and Eg2, I went through some datasheets... I think I found one that shows where my problem lies, below is the rp & gm characteristics for the 6BR7, but these characteristics should be applicable to all small-signal pentodes... You will note that rp increases when Eg2 is lowered, which makes more sense to me, since lowering the screen voltage lessens the electron flow, thereby increasing the internal resistance of the tube. Thoughts?

6BR7_rp.png
 
That's a nice chart, pity they didn't do that for all the tubes.

I only used the probe in the x1 position so it was 1M.

Not all small signal pentodes have the same internals but a good check is the g2 mu if listed, some are 20 and some are 28 or more which multiplies up of course.

You are right in that all pentodes will have similar shaped curves to the 6BR7 but the values will be different, there is no substitute for making a quick practical circuit with croc clips though, that removes all doubt.

best
DaveP
 
I think the discrepancy in my figures lies in the way the OPT is terminated and I'd like you to check my reasoning here.

Given that the OPT is 48k:600

Z ratio is 80:1, so the Voltage ratio is 8.944:1

As audio transformers are not 100% efficient and the actual loss is -20dB which is 10:1 making the efficiency 89.44%.

The OPT 600 ohm output is terminated with two 1.2k resistors making 2.4k all together.

When 2.4k is reflected back by the Z ratio it becomes 80 x 2.4k=192k.

When I draw this load line on the previous plate chart, the gain is 275Vp-p between -1.5V and -2.5Vp-p.

275 is  +48.8dB

Using this new figure for V3:-

IPT Gain +31dB
V1  Gain +25.9dB
V2  Gain +33.2dB
V3  Gain +48.8dB
V2/V3 fb -24dB
OPT Loss -20dB
Total Gain +94.9dB

Does this make sense?

best
DaveP

 
Good question:  did I terminate 600 ohms, or only have the built in pad? 
Any reference in the manual would be for 600 ohm termination, generally telco line expected. 
 
DaveP said:
there is no substitute for making a quick practical circuit with croc clips though, that removes all doubt.
I wish that was the case, unfortunately, the calculated rp's still do not look right... perhaps it will come to me later. ::)
 
There is a problem somewhere

The gain of V2/V3 in the closed feedback loop must be roughly equal to 1/B, where B=2.7k/682.7k=0.003954885.

So 1/B =252.8=48dB

Doug's figures in the closed loop are +24.5+34=58.5dB which is 843.6 which can't be right if the feedback loop was connected.

Help!

best
DaveP
 
Could be a fault in the unit I looked at, don't take it as gospel.  I didn't have it to check out for very long. 
 
First: WHY do you care about pentode plate resistance? In any practical *audio* amplifier the internal plate resistance will be far larger than the DC feed resistor or the load impedance.

> note that rp increases when Eg2 is lowered

But also Current is decreasing, thus Gm is decreasing.

Figure Vg2 does the same thing as Vg1, only less. The two together set the cathode current (Vp hardly matters above the knee).

I would suspect that, like a triode, Mu is fairly constant over a wide range of conditions.

In a simple triode, Mu can be measured with a ruler, simple geometry.

In a tet/pentode there are two geometries (G1 G2) in cascade to measure, and "small" deviations from perfection may multiply to real deviation. And we know plate-lines are generally not flat, but convex upward, so Rp must drop as Vp drops toward the knee. Still I think constant Mu is a fair assumption over most of the useful range.

So taking Dave's new figures, cathode impedance is 1.224uMho or 820 Ohms, Rp is 583K, Mu will tend to be 711. If you shift Vg1 *or* Vg2 to arrive at a different operating current, Gm will obviously change. For small change, Gm varies nearly as square-root of cathode current. (This is not the ideal theoretical result, but most commercial tubes follow the square-root trend well.) So if you run 2X the current, Gm will rise to 1.730uMho and Rp will fall to about 412K. But to get 2X the current with the same supply voltage, you have to cut the DC feed resistor in half. So tube Rp is even less significant.

If you are *really* asking how to maximize nominal voltage gain:

Run very low Vg2 and very low current with large plate feed resistor.

You find plate resistors as large as 1Meg in some DC regulator error amps.

We don't go that high in audio because stray wiring capacitance sucks-out the highs.

Also (and for the same reason) our loads are never >1Meg.

Note that the RCA uses one value at V1, which works alone, and a higher value at V2, which is inside a strong NFB loop which tends to correct a little droop at the high frequencies.

So pick the highest allowed plate resistor, considering load and strays. 

Mock it up.

Set Vg1 to zero (tube hard on).

Increase Vg2 until plate just comes off bottom. i.e. zero Vg1 is just enough to bottom the plate, for that supply and that plate resistor.

Now apply Vg1 until the plate is roughly halfway up the supply voltage. Don't go low, thinking high Ik means high Gm. The reduction in Rp will eat much of the apparent increase of Gm.

That will be nearly the highest voltage gain possible for that impedance level and supply voltage.

You may think that adding a buffer allows much higher impedance and thus gain. This is true; but most useful buffers can be used as amplifiers, and usually a 2-stage amp will beat 1 stage plus buffer.

Same argument for most active current "sources".

It should also be noted that two sections 12AX7 will beat any single pentode, for small added cost.
 
This got me thinking about my experience with modifying BA-1 and BA-11 NFB amounts, and this bit I found.

Digging in RDH3, on page 5, regarding pentode amplifiers:
1) plate R affects upper freq limit
4) miller effect of subsequent stage has a more pronounced effect than with a triode source
500K plate resistance given as having upper freq limit of 5KHz, generally speaking
 
PRR said:
First: WHY do you care about pentode plate resistance? <snip>
I don't really... :) the absolute value of rp is not so important but the direction in which it moves is...

If as you said, "Current is decreasing, thus Gm is decreasing." then rp should be increasing. But Dave showed just the opposite - with roughly the same Ep, Eg2 was lowered, and the rp decreased from "about 1M" in the datasheet example (Ia = 2mA), to ~580k from his calculation (Ia ~ 0.7mA). Since we know that mu stays pretty constant, the absolute figures for rp and gm could be glossed over a bit without causing too much trouble, at the end, it's the gain that we are concerned with. Anyway, I just wanted to bring this bit of oddity to your attention...
 

Latest posts

Back
Top