Harmon Kardon Citation 16 - UK Vintage Radio Repair and Restoration Discussion Forum

theredhouseinn · 29th May 2023, 4:42 pm

Got a problem chaps
One channel is ok.
I managed to make one good driver board from 3
So this board has all original transistors and components fitted.
The heat sinks have alternatives, 8 o/p BUX48A and the 2 drivers 2N3773.
These are the nearest alternatives that I could find, the RCA1BO4 and RCA413 no longer available.
In place of Q3 and Q8 I have used MPS6534, and Q4 and Q7 MPS6531, the 27722 and 25972 no longer available.
The problem I have is that the driver board is oscillating at 100kc/s. The first set of alternative transistors that I tried, the board oscillated at 20mc/s.
Using a variac the good board oscillates at +/- 30v then past 35v no osc. all the way up to +/- 70v.
So it would seem to me that the circuit is rather critical, even with the correct transistors in it.
The 4 transistors are termed general purpose transistors.
I am 65 years a service engineer, so I can fix the amp. if I have the correct parts. Knowing that the amp. did work before the power transistors gave up the ghost, and took a few other parts with them.
I cannot re-design a circuit to take into account the different characteristics of alternative parts.
If any of you guys are design engineers have a look at the cct. and say "ah yes" change this to that value etc., that is why it is hot.
John.

Radio Wrangler · 29th May 2023, 5:33 pm

There's nothing surprising there. A lot of transistor amplifier designs of that era were a bit close to the edge in terms of stability, not just in terms of the overall feedback loop, but also locally and in places where there isn't an overt loop drawn on the circuit diagram.

Transistors don't give as much gain as your average valve, stage-for-stage. This is well known. Also, they are a lot less linear and need more overall feedback or degeneration to linearise them. Also well known.

Less well known is that they pass signals BACKWARDS. With some attenuation for sure, but not always enough to compensate for their forwards gain.

The circuitry driving the input and the circuitry loading the output of a transistor stage can look reflective at higher than audio (getting on for RF) frequencies and you can get an oscillator circuit running with its signal passing through the transistor in BOTH directions and bouncing to and fro between the reflections at the source and load.

You don't see a loop or an oscillator on the circuit diagram, because we find it easier thinking of signals flowing through stages which only work in one direction.

A good many UHF oscillators have circuit diagrams which look like emitter followers.

Emitter followers are risky circuits for unintended oscillation.

The tail end of that Citation is packed full of transistors acting as emitter followers. Both the power devices on their heatsinks, and the smaller devices on the driver board.

It's likely that your replacement devices are significantly faster and gainier than the originals. Both of these factors conspire to make them friskier and the risk of oscillation goes up.

Even if you bought devices to the original types, it's likely that later production was friskier than the ones of the time the amp was designed.

The way out is to treat the devices as if they were wannabe RF devices and deploy things like "stopper" resistors which usually go in the bases, right close to the devices themselves.

For the power devices, something like an ohm or two in every device base connection would be a good start. Medium power devices might get 22 Ohms.

There are some signs that Harmon Kardon did some stopping with lead/lag networks in the area (series resistor-capacitor pairs)

Designing transistor amplifiers is not easy. Designing them to stay manufacturable and fixable as devices get "improved" is rather difficult. A slot of the data sheet specs which relate to parameters which affect stability margins are open ended with no real limit on max gain or max speed.

I think this one can be tamed at the cost of some time spent playing with stoppers.

David

theredhouseinn · 30th May 2023, 9:53 am

Thanks David.
Back in the old days we had grid stoppers on the valves to stop spurious oscillations.
I never twigged that the same could happen to transistor circuits.
I will start with the driver board, the heat sink is a major job.
John.

paulsherwin · 30th May 2023, 10:51 am

These amps were pretty bleeding edge at the time of their design, so transistor characteristics are likely to be much more critical than is typically the case.

29th May 2023, 5:33 pm	#2
Radio Wrangler Moderator Join Date: Mar 2012 Location: Fife, Scotland, UK. Posts: 22,874	Re: Harmon Kardon Citation 16 There's nothing surprising there. A lot of transistor amplifier designs of that era were a bit close to the edge in terms of stability, not just in terms of the overall feedback loop, but also locally and in places where there isn't an overt loop drawn on the circuit diagram. Transistors don't give as much gain as your average valve, stage-for-stage. This is well known. Also, they are a lot less linear and need more overall feedback or degeneration to linearise them. Also well known. Less well known is that they pass signals BACKWARDS. With some attenuation for sure, but not always enough to compensate for their forwards gain. The circuitry driving the input and the circuitry loading the output of a transistor stage can look reflective at higher than audio (getting on for RF) frequencies and you can get an oscillator circuit running with its signal passing through the transistor in BOTH directions and bouncing to and fro between the reflections at the source and load. You don't see a loop or an oscillator on the circuit diagram, because we find it easier thinking of signals flowing through stages which only work in one direction. A good many UHF oscillators have circuit diagrams which look like emitter followers. Emitter followers are risky circuits for unintended oscillation. The tail end of that Citation is packed full of transistors acting as emitter followers. Both the power devices on their heatsinks, and the smaller devices on the driver board. It's likely that your replacement devices are significantly faster and gainier than the originals. Both of these factors conspire to make them friskier and the risk of oscillation goes up. Even if you bought devices to the original types, it's likely that later production was friskier than the ones of the time the amp was designed. The way out is to treat the devices as if they were wannabe RF devices and deploy things like "stopper" resistors which usually go in the bases, right close to the devices themselves. For the power devices, something like an ohm or two in every device base connection would be a good start. Medium power devices might get 22 Ohms. There are some signs that Harmon Kardon did some stopping with lead/lag networks in the area (series resistor-capacitor pairs) Designing transistor amplifiers is not easy. Designing them to stay manufacturable and fixable as devices get "improved" is rather difficult. A slot of the data sheet specs which relate to parameters which affect stability margins are open ended with no real limit on max gain or max speed. I think this one can be tamed at the cost of some time spent playing with stoppers. David __________________ Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done

30th May 2023, 9:53 am	#3
theredhouseinn Hexode Join Date: Jan 2013 Location: Kelvedon, Essex, UK. Posts: 273	Re: Harmon Kardon Citation 16 Thanks David. Back in the old days we had grid stoppers on the valves to stop spurious oscillations. I never twigged that the same could happen to transistor circuits. I will start with the driver board, the heat sink is a major job. John.

30th May 2023, 10:51 am	#4
paulsherwin Moderator Join Date: Jun 2003 Location: Oxford, UK Posts: 27,946	Re: Harmon Kardon Citation 16 These amps were pretty bleeding edge at the time of their design, so transistor characteristics are likely to be much more critical than is typically the case.