This may be a problem that needs a wiring diagram or schematic and if so I can post one tonight but I'm hoping it's something that at least one or two people here have encountered before and will recognize.
In a nutshell, with pots for bias on the cathodes of single stage DHT's in both channels of a preamp, if I adjust the bias on one side, the bias of the tube on the other side tracks with it. Works both ways.
The B+ supply string to ground is PT > FullWave Bridge > LCLC > split to two channels >LC > Line Out Transformer > DHT > Cap bypassed bias pot > signal ground.
The Ground buss runs from chassis through the bridge > PS caps 1,2, 3 right/left. Each channel's signal ground buss running from first PS cap > bias pot > input volume pot wiper > input selector switch (grounds form inputs are switched with hots) .
(There is also hum and/or buzz that varies in sound and intensity with all the changes that I've tried so far looking for the fault.)
Present filament supply is on a separate board and is P Transformer > split to two separate channels consisting of FW Bridge > LCLC > LT1084 >
I've looked for shared ground faults to chassis but don't find any. Can there be a DC reference through the filament supply bridges referencing both filamentary cathodes to each other?
In a nutshell, with pots for bias on the cathodes of single stage DHT's in both channels of a preamp, if I adjust the bias on one side, the bias of the tube on the other side tracks with it. Works both ways.
The B+ supply string to ground is PT > FullWave Bridge > LCLC > split to two channels >LC > Line Out Transformer > DHT > Cap bypassed bias pot > signal ground.
The Ground buss runs from chassis through the bridge > PS caps 1,2, 3 right/left. Each channel's signal ground buss running from first PS cap > bias pot > input volume pot wiper > input selector switch (grounds form inputs are switched with hots) .
(There is also hum and/or buzz that varies in sound and intensity with all the changes that I've tried so far looking for the fault.)
Present filament supply is on a separate board and is P Transformer > split to two separate channels consisting of FW Bridge > LCLC > LT1084 >
I've looked for shared ground faults to chassis but don't find any. Can there be a DC reference through the filament supply bridges referencing both filamentary cathodes to each other?
Last edited:
Um, with no regulation of supply voltage any change of bias voltage (and consequently quiescent point / quiscent current) will inevitably change the B+ voltage of both tubes. Change in B+ will consequently lead to change in quiescent current of the other tube which (with resistive bias cathode arrangement such as potentiometer or straight resistor) will cause bias voltage to change.
Physics. Use fixed or didoe (LED etc.) bias if you don't like the way it works now.
Physics. Use fixed or didoe (LED etc.) bias if you don't like the way it works now.
Does the power transformer have independent filament windings for each channel's filament supply?? If not the bias is the sum of the plate currents flowing in both tubes - and adjusting the bias on one automatically changes it on the other as well. Sensitivity to the plate supply voltage should not normally result in more than a few % variation in the plate current of the other channel.
Thanks Arnulf. Yes I know that but in my experience so far if quiescent current draw on one tube is increased , that pulls down the B+ netting a drop on the other side . . . . sort of a See-Saw effect. In this case, when the current through one side is increased the current through the other side increases too.
Kevin, thanks and sorry. I wasn't sure how to paint the picture with words so it would be clear in one go. The power transformer for the filaments is an 18V Hammond feeding two separate rectifiers and their associated filter strings.
The B+ transformer is an old Lundahl 350VAC LL1650. I thought about using three of its 6 V windings for one of the filaments but asymmetrical loading of the Lundahl secondaries can sometimes lead to problematic hum fields, so I put two rectifiers and their circuits on the one Hammond.
I must be dense but I don't understand from this whether or not you have two totally separate filament windings (2 x 9V?) on your filament transformer - if you don't that is the issue, if you do then you have a wiring/design issue.
Can you post schematics showing clearly what you have done?
Hi Kevin, The Hammond only has one secondary - 18VAC. I'm using whatever I have on hand to explore a bit and so things are a little mis-matched.
I just tried putting one of the rectifiers on three of the Lundahl 6V windings and now the bias works as desired.
Thanks !
PS. Do you use buss grounding? I had the whole B+/Audio circuit on a single buss running from PS through everything with cathode and lastly grid at the farthest end. I also had hum that I thought might be associated with the cause of the bias problem and I tried to cure that by cutting the tube's ground circuit off the end of the buss and rerouting it straight to the first PS Cap. Now there's no hum but lots of 120Hz buzz. I'd have thought it would be a good idea. Perhaps not?
I do generally use buss grounding, and have no problems with hum or buzz in any of my components. (And this with 32Hz, 100dB+ efficient speakers)
Generally the first cap in my power supply gets the HT transformer center tap, the main chassis ground and the end of the ground buss, and that's it. Usually the highest current grounds circuits get placed near the last filter cap in the HT supply and everything follows from there.
The ground buss should be grounded at one, and only one point to avoid ground loops and to prevent supply ripple currents from developing a voltage across the IR drop of your ground buss.
Generally the first cap in my power supply gets the HT transformer center tap, the main chassis ground and the end of the ground buss, and that's it. Usually the highest current grounds circuits get placed near the last filter cap in the HT supply and everything follows from there.
The ground buss should be grounded at one, and only one point to avoid ground loops and to prevent supply ripple currents from developing a voltage across the IR drop of your ground buss.
That's interesting- in my boxes that use bus grounds, I attach the chassis ground at the other end of the bus, at the input.
OK, all is well- switched back to the previous buss scheme where it's essentially all in one row with grid at one end rectifier at the other . Because the way the bridge was physically situated I put chassis ground between rectifier and first PS cap. Now all hum and buzz is inaudible though for the first time I am hearing a lot of what seems to be tube rush but with a distinctly white noiseish shhhhhhhhhhhhh. I think I remember reading somewhere that this kind of noise can be caused by HF oscillation. I'll try something tomorrow but for now this is sounding pretty good. Thanks !
I tend to float my RCA jacks from the chassis, and I tend to also think a lot about ground loop issues caused by the jack grounds. If I am concerned about RFI I will sometimes install small disk ceramics between the RCA jack ground and the chassis. Outputs, particularly those that carry a lot of current tend to get grounded right at the "mecca" ground point (end of buss, 1st psu cap, trans CT, and chassis) I referred to in my previous post on buss grounding.
FWIW This approach echoes that used by the pro audio companies I worked for in the past, although in at least one instance I was responsible for doing it this way.
In the event of a power transformer insulation breakdown this provides an extremely direct solid path for fault currents to flow directly into the safety ground and not through the entire grounding architecture of a PCB which may not be able to support the fault current, thereby possibly opening the trace to the chassis gnd connection before the fuse blows. This a potentially lethal combination with PCBs..
I was also taught by my audio design "mentor" long ago to place the chassis ground connection at the noisiest place in the power supply, and it has worked extraordinarily well for me. Also it is really important that the loop carrying the input capacitor charging currents goes directly and only to that point. (Whether transformer CT or the 'return' leg of a bridge rectifier) In any design there really should be just one place where all circuit grounds are referenced to and with multiple supplies/analog and digital grounds it just usually makes sense IMHO to place all of the supply commons together at this point. I then also make a low impedance connection between this point and the chassis.. (I may also at times 'carefully' use a high current ground lift [anti-parallel high current diodes in parallel with a low ohm power resistor] at this point to break ground loops between the safety grounds in multiple components.) The safety ground is connected directly to the chassis at the IEC input connector in all of my components.
Last edited:
My reading on the buss ground scheme is that it boils down to a single point ground where the "point" consists of a very low impedance (at audio frequencies anyway) single conductor which optimally is arranged with the high current returns at a single point and only circuit returns with minimal current connected to the conductor at any distance from that point. Imagine a true star ground with a bundle of wires from say the preamp section. Replace that bundle with a single conductor having taps along it's length.
If the impedance is very low and circuit currents are likewise very low, the unintended coupling should be insignificant. If only audio signal current returns are placed on the ground buss (nothing with noise or ripple current) then it seems to me that going to the trouble of a discrete star ground may be unnecessary. There is still the question of the buss acting as an antenna or a resonator at e.g. cell phone frequencies...
Placing the ground star near the higher noise current sources makes a lot of sense to minimize the loop area of the physical circuits carrying these currents.
If the impedance is very low and circuit currents are likewise very low, the unintended coupling should be insignificant. If only audio signal current returns are placed on the ground buss (nothing with noise or ripple current) then it seems to me that going to the trouble of a discrete star ground may be unnecessary. There is still the question of the buss acting as an antenna or a resonator at e.g. cell phone frequencies...
Placing the ground star near the higher noise current sources makes a lot of sense to minimize the loop area of the physical circuits carrying these currents.
I'm thinking about the magnetic loop formed by the wires conducting for example the power supply filter return current. If these conductors go to a star ground 12 inches away, there is a 12 inch long partially open "single turn" creating a magnetic noise field. One could always twist the transformer CT or bridge (-) wire together with the return from the filter cap going to a remote star ground, but to me it's easier to place the star point relatively close to the power supply and have the power supply current and audio returns fan out in opposite directions. I usually put the power supply at the "back", signal path in the "front" with the star ground situated between.
Attachments
So when you build a multi-section power supply filter do you run a separate ground return to each of the filter capacitors?
No, to keep the rectifier switching current loop short, the CT of the transformer or bridge return connects to a terminal on the cap of the first stage, then to the second filter cap, etc, daisy-chain fashion to the star ground. The filter currents are all local and there is no ground loop created by the daisy chain connection because the transformer secondary is isolated and the filter cap (-) terminals are also isolated.
I guess my description was a bit ambiguous; I should have said that the end of the filter return wiring chain which starts at the CT etc. terminates at the star point, rather than the CT and all the separate filter cap returns wired separately to the star point.
In stacked supply amps, I wire both filters this way and have a second isolated star point for the connection between the 2 supplies in addition to the ground star point.
Last edited:
OK thanks Michael, I see what you're saying now. It's helpful.
I have another question for you on an unrelated topic. While we're talking I'll ask you here. If it turns into a thread maybe it can be moved. (?)
I have two pairs of PP outputs here I'd always thought I'd never use until I started reading your posts on the anti-triode. My question is. If you take it that each plate of a pair of triodes in push pull sees 1/4 of the primary's impedance, what does the triode's plate see when its counterpart is an anti-triode delivering complimentary current?
I have another question for you on an unrelated topic. While we're talking I'll ask you here. If it turns into a thread maybe it can be moved. (?)
I have two pairs of PP outputs here I'd always thought I'd never use until I started reading your posts on the anti-triode. My question is. If you take it that each plate of a pair of triodes in push pull sees 1/4 of the primary's impedance, what does the triode's plate see when its counterpart is an anti-triode delivering complimentary current?
Last edited:
- Status
- Not open for further replies.