• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

audio innovations 200

My problem is as follows. I own an Audio Innovations 200 power amplifier. It consists of half a Ecc82 amplifying device, then the actual power amplifier follows: an ecc88 paraphase, followed by a push-pull el84 arrangement. Overall feed-back is taken from the output on the secondary side of the outputtransformer to the cathode of the ecc82.

By accidence I discovered that the operating point (voltage and current) of the ecc88 is only correct when there is a speaker attached to the amplifier. Without speaker one half of the ecc88 draws less current, therefore operates on a higher voltage on the anode and this in turn affects the current drawn by one of the el84's (less current). I built myself a sort of a copy of this amplifier and in my case all setting are correct, speaker attached or not.

My questions:

1. Is this problem evident in all AI200's or is it just one strange case?
2. What can be the reason for this abberation? I have checked all components, C's, R's, tubes, transformers, nothing is wrong. Both channels display the same abberations. I can think of nothing except a serious oscillation without load that causes DC-shifts?
3. How can the problem be solved?

By the way, when used with a speaker this amplifier works very fine and sounds beautyfully.
This is one of those situations that has Something-Is-Wrong written all over it. Unfortunately, it's not going to be easy to diagnose long distance.
Tube amps are notoriously stable. You can operate them open loop, closed loop, and upside-down--they don't care. I think it's safe to say that the bias should remain constant regardless of whether there's a load or not.
Your guess as to an oscillation is a possibility. You'll need an oscilloscope to check that.
Is the second stage (phase splitter) direct coupled or capacitor coupled to the first stage? Check to see if the bias to the phase splitter (both sides) changes when the load is hooked up. You say the current changes, but let's get more detailed. If the phase splitter is cap coupled, the grid should be within a few millivolts (DC) of ground. If the bias changes, you may have a leaky DC blocking cap between first and second stages. This might test as good on a cap meter, but pass DC under high voltages.
It's also bothersome that the output tube changes. Is the phase splitter direct coupled to the output stage? How is output stage bias handled?

Thank for the reaction. The stages are C-coupled and they're not leaking. After the first stage and cap however the second half of the paraphase is driven (signal applied to the grid, attached to ground via 100 k resistor). Indeed somehow here is the result a -4 Volts which passes to the grids of the output tubes. It is not the grid that is drawing current however.

If I take the ecc82 (very first stage) out of the amplifier (and effectively make the amplifier open loop) all problems are gone. It has to be a oscillation (I have no scope). But if this is true, all AI200 must have this problem!

I do whenever I like. Never had any problems with it. It this part of a myth that under no circumstances should any tube amplifier be turned on without a load? What is the worst thing that could happen? A tube or 2 that die? New ones in it and there you go. From the same school that says tubes should be replaced every few years?
Not a myth. An all too common reality. The usual result is a blown OPT (expensive); the valves (cheap) usually survive. Never run a valve amp without a load. In your case you are lucky and have got away with (so far). All that is happening is that your amp is oscillating at some ultrasonic frequency. Dogs or bats in the vicinity might find this unpleasant, as the OPT laminations are probably singing due to magnetoconstriction.
I understand your point and that I had some luck in the past. Didn't know that. But shouldn't the amp be designed in such a way that this outcome is highly improbable (in other words: an amplifier that gets into oscillation that easy is not an example of good engineering practice). I'm just curious how many Quad II's have died during 40 years of use because of 'not putting a load on it' (of course not being fed with a signal at the input!).
Adding a Zobel network at the output will probably stop oscillation, and reduce the risk of OPT damage. Traditionally these have not been fitted to valve amps as it was assumed that users would read and follow instructions. In the 1950s it was less necessary to make everything 'idiot proof' because in those days if someone did something stupid and harm resulted they would blame themselves, rather than looking for someone else to sue.
its fine, these amps are design with having speakers attached. otherwise, the circuit would be too dependent on the resistance of the output transformer's DCR. and the lower the DCR with high inductance the more of an offset would be observed when the output transformer is not loaded.

reducing the tube noise impressed on the transformer helps as this will load the tube in the RF region. take a 1 pf cap across the output transformer plate to plate connection.