I've designed and built a fully-balanced line stage for a preamplifier based on a long-tailed pair of E88CCs / 6922s with a cascade constant current source in the tail and cascaded current mirror active loads. I've used a similar configuration in the past, which has worked very well, but this particular circuit generates about 150mV of common-mode noise on its outputs! I have tried numerous things to work out the cause of the problem but have so far failed to fix it.
The tail current is set at about 11mA and the active loads are configured to set the anode/plate voltage at around 80V. The +200V and -15V supplies are regulated and completely quiet, and there are 100nF decoupling capacitors to the ground plane where the supply rails connect to the circuit.
The fact that the noise is common mode, first suggested to me that the problem lay in the tail current source. However, The voltage across the 220R resistor is completely stable and quiet. I then turned my attention to the current mirror, but the voltage on the emitters of the upper transistors is also completely stable.
This has me completely baffled. Can any of you brainier guys tell me where I'm going wrong?
The tail current is set at about 11mA and the active loads are configured to set the anode/plate voltage at around 80V. The +200V and -15V supplies are regulated and completely quiet, and there are 100nF decoupling capacitors to the ground plane where the supply rails connect to the circuit.
The fact that the noise is common mode, first suggested to me that the problem lay in the tail current source. However, The voltage across the 220R resistor is completely stable and quiet. I then turned my attention to the current mirror, but the voltage on the emitters of the upper transistors is also completely stable.
This has me completely baffled. Can any of you brainier guys tell me where I'm going wrong?
Yes, in effect the anode loads are separate current sources with a common reference voltage. This is a handy circuit that I came up with a while ago that enables the gain for the stage to be maximised and accurate current balance to be maintained. It's worked very nicely in other designs that I have done, which is another reason why I'm baffled.
Yes, the component values are much the same as I have used before. When you refer to a high resistor value, are you talking about in the tail current source or in the active loads?
As for the type of noise, I'm not entirely sure. I guess I may need to feed it into a spectrum analyser to find out - or is there an easier way?
The 10uF capacitor in the active load reference voltage network keeps the programmed current substantially constant across the audio frequency range and also decouples noise from the bases of the upper transistors.
As for the type of noise, I'm not entirely sure. I guess I may need to feed it into a spectrum analyser to find out - or is there an easier way?
The 10uF capacitor in the active load reference voltage network keeps the programmed current substantially constant across the audio frequency range and also decouples noise from the bases of the upper transistors.
I shorted both grids to ground and attached scope probes to each anode. With the scope inputs AC coupled I was able to assess the peak-to-peak noise signal to be around 150mV. Using the maths feature on the scope, I can subtract one input from the other, and the result is noise-free. If I feed a balanced sine wave signal into the grids, I get very noisy sine wave outputs on the anodes. Again, if I use the maths feature on the scope, the result is a clean sine wave (okay, there's a bit on noise on it as one might expect because of CMRR).
There is noise on the cathode that mirrors the noise on the anodes, albeit at a much lower level as one might expect. What I'm not clear about is whether this indicates that the noise is originating in the current source or elsewhere - after all, I would expect to see the noise signal on the cathode given that it is common mode. I suppose I could try replacing the current source with the fixed resistor and see if the noise is still there. I can't see why my current source should be producing noise, though...
I had thought about this but hadn't thought it to be the cause because the noise signal didn't look like an oscillation. I've used 1K - and even 470R - grid stoppers with E88CCs in the past with no oscillation problems.
Whilst I can't rule parasitic oscillation out based on what I've observed and measured, I would have thought it would result in a differential mode signal rather than common mode. Or am I wrong...?
Whilst I can't rule parasitic oscillation out based on what I've observed and measured, I would have thought it would result in a differential mode signal rather than common mode. Or am I wrong...?
It's a stereo preamp, so I've built two channels. They both exhibit the same problem, so I think we can probably rule out a faulty component. Just to be on the safe side, I have tried replacing all of the current source transistors but it didn't make any difference. I've checked for any build faults and have found none.
Well mentioned..I had exactly the same dilemma; found out the heater AC current was slightly vibrating the mica cathode/ grid supports producing a perfect tuning fork 5kHz tone...til one knocked the tube and it disappeared. Fair to mention, I found many variants of the mass produced US 6922 NOS are particularly prone to this.
richy
It's important that it be filtered for common-mode, not just differential. If the heater transformer winding is on the same core as the B+, common-mode noise can be very efficiently coupled since the cathode impedance is high. I wouldn't rate this as the most likely culprit (I'd suspect the cathode CCS), but it is possible, even with a regulated heater supply.
Thanks for all the helpful suggestions. I will pursue the RF oscillation issue in more depth because this seems like the best lead to follow at this point.
I'll try increasing the value of the grid stoppers and see if that helps. I'll also look again at the heater supplies. As I mentioned, these are regulated and elevated to 50V; and each heater terminal is decoupled to 0V by a 100nF capacitor placed as close as possible to the valve pins. Perhaps a common-mode choke connected to the heater pins and placed close to the valves would also be worthwhile.
I think the active load circuit should get around any current hogging effects because it matches the current in each valve section to within 10uA or so.
Are there any other tricks I could be trying to stop any RF oscillation?
I'll try increasing the value of the grid stoppers and see if that helps. I'll also look again at the heater supplies. As I mentioned, these are regulated and elevated to 50V; and each heater terminal is decoupled to 0V by a 100nF capacitor placed as close as possible to the valve pins. Perhaps a common-mode choke connected to the heater pins and placed close to the valves would also be worthwhile.
I think the active load circuit should get around any current hogging effects because it matches the current in each valve section to within 10uA or so.
Are there any other tricks I could be trying to stop any RF oscillation?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.