AVO 163 amp board ~ transformers

[Dekatron]

Originally Posted by Bazz4CQJ

OK Martin, so would you say that the right approach would be for Ed to consider whether he can duplicate the original transformers as closely as possible?
B

As a first step that would possibly help some people who have said that they have open circuit transformers. But apart from HBWOODY I don't know of anyone who has this problem right now, most people I've helped have just had to replace the electrolytic capacitors, some resistors and in a few cases a transistor or two to get their amplifier/oscillator to work again.

Then when we know that these transformers work properly we can continue the new design, and we can reuse the transformers in the new design.

So if anyone who reads this has a damaged oscillator transformer who he could send to Ed that would be very helpful.

[Dekatron]

Another problem with the amplifier board is that the 1.99µF Styroflex capacitor sometimes go open circuit.

What should we replace that capacitor with, I haven't been able to find someone who manufactures these types of capacitors nowadays. I have replaced them with Polypropylene capacitors and that works well enough but I am not sure if that's the best type to use.

[Radio Wrangler]

Ah-HA.

Styroflex....

Another piece of the jigsaw slips into place.

That ferrite core... N28 or N48 material (Siemens originally, then Siemens-Matsushita, now Siemens-TDK) just happens to have a temperature coefficeint which cancels that of polystyrene capacitors. So the tuned frequency will tend to stay fixed over temperature.

I suspect the Vinkor series core was of a comparable ferrite, if someone at AVO knew that trick.

David

[Bazz4CQJ]

Originally Posted by Dekatron

As a first step that would possibly help some people who have said that they have open circuit transformers. But apart from HBWOODY I don't know of anyone who has this problem right now, most people I've helped have just had to replace the electrolytic capacitors, some resistors and in a few cases a transistor or two to get their amplifier/oscillator to work again.

I think you may be right about the amp transformer; the HBWOODY situation is unusual.

But as you then say, you have replaced caps, resistors and transistor to repair these boards (I've replaced caps and resistors) and this is why I want to pension them off. But let's collaborate on our different strategies; I'm sure that would be worthwhile! PM sent.

I suspect that Avo needed to achieve temperature stability to suit various climates and situations where the testers were in use all day. I'm not sure that any changes we make absolutely need to equal that requirement, but it would be good if we could.

B

[Ed_Dinning]

Hi Folks, a little investigating on HBWoody's board
The transformer is reading 93uH /0.4R the pri and 992mH/ 40R on the sec, all measured on a TF2700 bridge, with sensible Q figures on loss balance.

The anti parallel diodes on the sec showed about a 30% imbalance between forward drops as measured on an AVO resistance range

The Styroflex cap measured at 1.96uF, no leakage

Looking at the circuit the transistors are spec'd as various gain grades of 2N2926's. BC549's are fitted

From the Towers book we have VT1 must have a gain >235 @2mA
VT2 gain >55 @2mA
VT3 gain >90 @2mA

BC549's can go as low as 110 @ 2mA.
The towers given equivalent is BC109, which can go down to 120 @ 2mA; so something to think about there!

Ed

[Dekatron]

Ed, Are you sure that you didn't look at 2N2925, that one has a minimum hfe of 235 where I looked but the 2N2926 series had lower values.

Also my databook says that a BC549C has a minimum hfe of 420 at 2mA, but at 10µA it falls to an hfe of 100.

I did som Googling on this and found that different manufacturers give different figures, NXP/Philips have the figures I quoted but ON Semiconductor has the lower hfe figure at 2mA! I've never had any problems with replacing any of the 2N2926 with BC549C (or BC547C) but I've never bought any from ON Semiconductor, only from National Semiconductor and Philips/NXP. So we might have to be careful in what brand we use and check them before use.

A gentleman in Sweden just replaced his 2N2926, all of them on both oscillator and amplifier with BC549C, I don't know of which brand though, but it worked just as it should after the change.

[Radio Wrangler]

2N5087?

David

[Dekatron]

2N5087 is a PNP but we need NPN here.

[Ed_Dinning]

Hi Martin, my Towers is the second update, about 1980
It lists various 2N2926 types, from an unsorted device to those with dash numbers as well as the various colour spots

The BC 549's fitted are all unsorted devices, again I took the Towers unsorted device parameters for my mail above

As you say, different manufacturers devices may have different figures. If I get a chance I'll take them out and measure them.
Certainly the schematic is specific about which type can be fitted where an its min hfe value

Ed

[Bazz4CQJ]

Thanks for reporting that valuable info Ed .

B

[Dekatron]

Thanks for the info Ed, it would be nice to if you could measure the ones fitted.

[Ed_Dinning]

Hi Martin, just did a quick measure on the VT1, which needs high gain, using a B&O tester it shows as 400, but I have no way of knowing the conditions other than "low power"

Ed

[Dekatron]

That seems like the figures I got on the few Philips/NXP that I tested, but I also got a few that was way higher, almost double that figure.

It seems like the only way to check the amplifier is to hook it up to an oscillator with a variable output that can be varied down to the mV range and test it, using a DMM on the µA range across a 1500 Ohm resistor in place of the meter outputs to see how it behaves over that range.

Checking all components might be necessary too.

[Bazz4CQJ]

Might 2N3904's fit the bill; certainly the pin-outs are the same and I'm sure the gain is typically pretty good? I have a bag full; I'll dig them out.

B

[Craig Sawyers]

I'm not sure why transistor hFE is important.

The first stage is an emitter follower, whose gain is hfe/(1+hfe). Note that most transistor testers measure static hFE at a particular current, but the gain of an emitter follower is dependent on the incremental hfe. These are not so different, but even if the current gain is only 100, the gain of the emitter follower is 0.99. The native input resistance is quite high, around 500k, but it is actually dominated by the parallel combination of the base bias resistors to 10k.

VT2 gain is set by RC/RE to 3x

VT3 is set by RC/re because the emitter resistor is bypassed; re = 25/IE with IE in mA. IE in this case is 0.3mA, so re ~ 80 ohms so the stage gain is ~ 275.

The overall gain of the amp is ~ 275x3 = 825, independent of transistor gain.

Then the actual gain is set by feedback from the voltage across the meter flowing via C8 and C9 into the 200 ohm resistor R14, fed back into the emitter of VT2. Looking into the emitter the impedance is not the 3.3k resistor R9, but our old friend re (in parallel with R9), which for VT2 is 40 ohms. So the overall gain is (I think I have this right) 200/40 = 5x.

But whatever - the native gain and fed back gain is hardly dependent on transisor hfe at all. HFE mainly impacts the base voltage (and hence emitter current) a little bit because the base takes a bit of current from the potential divider that sets the operating current. But that is a small effect for any reasonable current gain

So I'm not certain at all why AVO specified gain bands for the three transistors.

Craig

[Dekatron]

The original 2N2926 were sold with different gains, hence their colour coding, and the price varied with the colour code. So it was just a way of testing the transistors during manufacture and then selling them according to what hFE they had I guess.

To get the circuit to work properly AVO had to select hFE properly, therefore they specified what colour each transistor should have. There is one change note where they specified that VT3 could be any except brown, so they were aware of the problems with incorrect hFE in some positions.

[Dekatron]

Here's the hFE table for different colors of 2N2926.

[Craig Sawyers]

I've stuck the raw amplifier circuit (without transformer and feedback) and the overall gain is 316.

Gain of the first stage is indeed 0.99

Gain of the third stage is 243, very close to the predicted 275.

The second stage gain is lower than I expected at 1.33. What I had not taken into account is the output resistance of the second stage is Rc = 10k, and the input resistance of the bias network of the third stage is 82k//20k = 16k. So the gain is reduced to 2x. Then it is further loaded by the input resistance of the third stage amp, which will be hfe x re. And that is why they chose the highest hfe grade for VT3

Substuting a BC550C with an hfe of 450-900 increases the amp overall gain to 333. The loading from T3 is still significant though. With an re of 80 ohms and an hfe of 400 the input resistance of the transistor is 32k. So the loading on the collector is 82k//20k//32k = 10k. so the expected gain of T2 is 3.3 x (10/(10+10) = 1.65. Spice gives 1.5 which now agrees well across the board.

Craig

[Dekatron]

Originally Posted by Craig Sawyers

Spice gives 1.5 which now agrees well across the board.

Craig

Could you share the Spice files? I might learn something from them. I use LTSpice and hopefully I could get your files to work there if you used anything else than LTSpice.

[David Simpson]

Looking from the outside - in, guys have done sterling work in transformer theory & transistor theory - to the n'th degree. Plus Ed's great work.
Chronologically, along comes the MK3's & 4's, which are then shrunk & morphed into CT160's - at about the same time as the CT446 Tr. tester, and finally the VCM163. No doubt the AVO Factory engineers & technicians would've used their newish & faithful CT446's to carry out tests on the transistors used in the 163 along with the parameters in the newish "Transistor Data Manual".
Being - - hFE (or dashed beta) = DC or large Signal Current Gain, beta( or hfe or alpha dash) = small signal current gain. And I'co (or I CEO) = leakage current. This is all back in the 1960's.
What is needed now is a means to keep 163's fully functional in the 2020's. Especially as decent fully restored & calibrated ones cant fetch £1000 or more on the open market. More critically - right now - folk need to know what phooking procedure was HBW undertaking when his Amplifier Board failed ? ? Thus ensuring that other 163 owners(particularly inexperienced ones), don't accidentally commit the same sin.
Bazz's idea of a modern PCB, or a modified one is maybe the way forward. But quite frankly - the original circuit boards look jolly robust.

Regards, David