RCA OP-6 Analysis

[DaveP]

Right, it's time to settle the gain issue around V2/V3.

Beta is almost the same for early and late models.

Early= 2.2k/562.2k=0.003913 so 1/B =255.54

Late=2.7k/682.7k=0.0039548 so 1/B=252.85

These figures are in close agreement and they represent the maximum gain of V2/V3 if the gain were infinite or like an op amp.

Because the tubes have a finite gain, the combined gain of the tubes with feedback can never be more than ~250 or  48dB.

This means that Doug's figures are not correct for a perfectly functioning OP-6 or he made an error, we shall never know

What we do know is that the gain of V2 without feedback is 46  (33.2dB)  and the gain of V3 could be between 80 and 192 depending on the reflected load. This gives us a maximum and a minimum combined gain of 3680 or 8832, so let's see what that does with feedback.

Using the Beta from the later model, the Gain closed loop is  236.6 for the minimum and 245.8 for the maximum.

This means that V2/V3 gain is between  47.48dB and  47.81dB, hardly any difference at all.

With this settled, we can move on to V1 and the feedback there.

Best
DaveP

 

[emrr]

I can only guess the NFB had been monkeyed with in the unit I measured.  Speaking of monkeys, here's Tarzan with an OP-6.

 

[CJ]

would be interested in DCR and freq graph of I/O iron,

 

[PRR]

> rp should be increasing. But Dave showed just the opposite

With one sample on the datasheet (which may be wrong) and one sample in Dave's hand (which my be atypical).

If it is important, you should probably set up your own test jig.

 

[jazbo8]

> rp should be increasing. But Dave showed just the opposite

With one sample on the datasheet (which may be wrong) and one sample in Dave's hand (which my be atypical).

If it is important, you should probably set up your own test jig.

Nah, there is no need... I should have thought of it earlier - just look at the handy RCA Conversion Factor Chart below, check out the Fr line, it clearly shows the lower the Vg2, the higher the rp; and the lower the Vg2, the lower the gm - exactly as what one would expect.

 

[DaveP]

Moving on to V1.

I've tested it without feedback to get a benchmark.  This was done with the probe on the x10 setting with no load and it gave a gain of 170.  With the probe on the x1 setting the gain is exactly 140 and with a 1M load the gain is 122.5. (Jazbo8 please note!)

Using the standard feedback formula, the Gain closed loop = 170/(1+(170x47k/1247k)),  =22.95, when I measured it with probe on x10 with no load, the gain is 21.5.  This is equivalent to  27.2dB and  26.6dB, so that checks out close enough.

This circuit is actually quite remarkable, I measured a maximum output of 180Vp-p before clipping,  which is 63.6V RMS, so the headroom is amazing.  With a 1M load connected, the output falls to 175Vp-p which means the output impedance is ~28k.

This circuit has many merits  and the 17dB of feedback is obviously beneficial, the only demerit is that there is a 47k resistor in the grid circuit.  This will generate 2.75uV of noise in a 10k bandwidth and if you followed it up with 59dB of gain (90dB less IPT) you will have 2.4mV of noise at the output.

Nice small signal pentodes have an additional noise of ~2uV but noise voltages are not totally additive as there is always some anti-phase cancellation going on, so if we say the total noise will be 4.3uV then the output noise will then be 3.876mV.  As the max output is +19dBm or 6.9V this would give a signal to noise ratio of 65dB, very respectable for a 1940  portable pentode pre-amp.

The gain control is a very interesting ladder attenuator that is no longer made by the manufacturer.  It enables the gain range to be extended by controlling the amount of feedback and the load simultaneously, I think it's very smart.  I understand that they can be obtained from EMRR if you ask him nicely.

All in all, I think it's a very well thought out mic pre and it may be that the transformers and the unique circuit of V1 give it some special magic, that is something I can't test or measure.  Having said that, if I had to make a high gain tube mic pre for a ribbon mic, I would probably use a V76 front end, but that's another topic   This is a fraction of the work that some of us do before starting a project, unless you understand what you are going to make before you make it, you will fall at the first snag.

I hope you have enjoyed this thread as much as I have doing it and have learned as much as I have too, many thanks to all of you who have contributed.

Best
DaveP

 

[CJ]

easy circuit to build if you want to do a transfer function plot,

note that 400H choke was  probably checked at 20 Hz, it will have much lower Henries  at 1K Hz,

transformers are usually about 99 percent efficient, so loss should not be a factor for calculations,

note the .25 cap feeding the OPT, V76 uses 4 uf, so you may notice some bass loss as .25 uf = about 30K at 20 Hz, if OPT is 250 H pri, then XL = 31K at 20 Hz so you have a 50/50 voltage divider,
may be limited on purpose for  AM radio, V76 has freq roll off on input circuit,
or maybe they do not want to saturate the core,

V76 uses same feedback/attn gain switch concept, 

 

[gridcurrent]

note that 400H choke was  probably checked at 20 Hz, it will have much lower Henries  at 1K Hz

output transformer K-900787-501
terminals 4:2 (inductor)
4300 ohms DCR
400 H @ 1000 Hz  (Q= 0.4)
427 H @ 120 Hz (Q=10)

 

[emrr]

hold da phone:

Where did 400H come from?  Did I say that?  I don't think I did. 

Eyes on a manual with spec reads 200H for the choke portion. 

The 900787-2 example I measured IN CIRCUIT with an Extech 380193 L meter gave:
92H @ 120, 116K Z
279H @ 1K, 130K Z

The output primary portion read:
92H @ 120, 121K Z
264H @ 1K, 130K Z


An earlier version 900787 measured 4030R for the choke portion and 1414R for the primary. 

 

[lassoharp]

I've seen that same behavior for some transformers when using handheld meter.  Usually the L at 1K as CJ said, is much lower than it will be at lower freq.  But for whatever reason, some iron gives a higher reading at 1K than at 120Hz

 

[gridcurrent]

yet a different OP-6, serial number 18xx.
measured with a Tenma 72-960 LCR meter.
tested against a UTC decade inductor box model DI-1
position 10 marked 100.5 mH,  the meter reads 98.1 mH @ 1000 Hz .

output transformer K-900787-501
terminals 4:2 (inductor)
4420 ohms DCR
480 H @ 1000 Hz  (Q= 0.4)
416 H @ 120 Hz (Q=13.8)


there is a OP-6 floating around  the southeastern  US with a kludged output transformer.
the inductor's measurements were as follows:
59.7 H @ 1000 Hz  (Q=3.95)
50.3 @ 120 Hz (Q=7.22)

 

[emrr]

Thanks.  I always assume a -501/2/3 part # is the later version over a (-nothing), -1, -2. 

It is funny how much L variance can be found amongst the same part, even same run in the same equipment in the case of multiple channels. 

My Extech meter matches up closely to  a decade inductor, like yours, for both 120 and 1K. 

Anyway, the one manual spec I find in print says 200H, to be clear. 

 

[CJ]

remember that the choke has DC current running thru it while in circuit, this will lower inductance,

also, hand held meters use a very low AC voltage to measure inductance, in circuit AC voltages will be much higher,

laminations start to give out as f ramps up to  around 1000 Hz, that is why L drops so fast, but freq multiples reactance, so XL is still high,

here is an in circuit test on a V76 choke>

 

[gridcurrent]

remember that the choke has DC current running thru it while in circuit, this will lower inductance,
also, hand held meters use a very low AC voltage to measure inductance, in circuit AC voltages will be much higher,

The Tenma 72-960  LCR meter outputs approximately 100 mV while testing an inductor.
What would be a better, reasonably priced test setup ?
General Radio 1650 bridge (or similar) with an external oscillator ?

 

[CJ]

how about a bridge where you can control all the variables like voltage and frequency?

for zero $$$?  (if you have the lab equipment already,    )

inductors are just AC resistors,

so you can use ohm's law for inductors, which is Ohms = V/A

now take it into inductor world,  Ohms = 6.28 f H,  we want the H,  so isolate with algebra,

H=Ohms/6.28 f       

there will be errors from DCR and core loss, DCR error will show up at 5 Hertz where XL (ohms) approaches DCR to a noticeable extent, depending on what kind of coil you are working with,

core loss can usually be ignored as audio stuff is small, as loss is proportional to weight,

the good thing about this method is it will give you the same answer every time,

we used to use a BK precision, Sencore, and Gen Rad bridge for taking measurements, used to get 4 different answers, if we did not exceed the Henry value for the Sencore which was rather low,

that funky Gen Rad has a Series or parallel switch that gives two answers for the price of one,

just be sure that you do  not exceed the saturation point of the coil.core you are checking or you numbers will go haywire,

we are using this setup tonight on the 9043 and 9050 Neve eq inductors, which is nice as you can check individual taps at the freq where they are being used,

anybody know what voltage level these guys see?  ???

 

[DaveP]

This topic is nearly four years old, but I just discovered something of interest.

To recap, this portable mic-pre has aquired fame and a very high price, but it was not designed for studio work, mainly outside broadcast.  I attempted some analysis back in the day, but I omitted to look at the noise spec.  I have attached the spec again for reference.

At the output level of +18dBm  (6.2Vrms) the noise level is -70dB, this means it is at  -52dBm, but this is at a gain selection of +68dB.  This means that the self noise or EIN is 68dB below -52dBm which is -120dBm!!!  This is V76 territory!

This has changed my view of the design and it shows just how good the RCA guys were back then and that they managed to get that spec out of 1620 tubes from the 1930's, tubes with exposed grid connections.

It may well be possible to tweak this design for the better after 70 years, especially since ribbon mics have a higher output nowadays and that the most gain anyone might need for say a SM7 would be 60-70dB, not 88dB.

Food for thought........mmmmmmmmm

DaveP

 

[emrr]

Hum has always dominated hiss in all the units I’ve seen.  Using the battery connection for an external DC supply rather than  internal AC gives about a 12dB hum improvement, this in my opinion makes them actually useful, and I suspect battery power  is how that spec was procured.  Too bad most people ditch the stock power connector for an IEC AC inlet when a modern restoration is done.   

The grid is never more exposed than any other tube, if protocol is followed.  Shielded grid lead and shield cap on the tube. 

 

[DaveP]

Can anyone throw any light on why one end of the OPT primary is connected to the cathode of V3 rather than ground?

The cathode resistor is not by-passed, does this type of connection get around the resulting higher internal resistance?

DaveP

 

[emrr]

Can anyone throw any light on why one end of the OPT primary is connected to the cathode of V3 rather than ground?

The cathode resistor is not by-passed, does this type of connection get around the resulting higher internal resistance?

DaveP

In the old books, that's mentioned as a quieter path connection.  Slightly higher gain too?  Can't remember about that. 

The other thing with OP-6 is the input transformer is 27-28dB gain, which is unavailable. 

 

[volker]

Sowter sells the V72 input transformer, which can be connected as 1:20.