Thanx alujoe2 for including a schematic with each plot - that really is so much easier to then do a simple cross-comparison for any viewer and allow a directed comment for each post.
Yes schem #6 is the step I consider worth making and ploting, as it routes the final bias cap C16 directly to the node used by the two cathode sense resistors and the B+ supply to the OT. That node is the appropriate common node for Vgk of each KT88, and the output stage B+.
Thanx for elaborating on the 650kHz issue. As 6A3sUMMER indicates, there may well be a problem with the B+ feed to the OT. Any OT primary signal current has to loop through the KT88's, the common cathode node, the last B+ filter cap, and back to the B+ feed on the OT. That loop contains the highest signal current level, and is imho, the most important loop to minimise for loop area and stray parasitic capacitance and inductance.
You indicate that common cathode node has a long lead run to the last B+ filter cap neg terminal, and that lead length can't really be reduced. Imho, the two cathode sense resistors need to separately run to the B+ cap negative, and that negative terminal should be a star for the bias 0V connection, and the 0V link to C2 and preamp circuitry, and the OT secondary winding ground. That star can have a link to chassis. The PE blocker circuit should go independently to chassis as it is effectively the PE connection to chassis, and has to handle any high current protective function.
The topic of the protective relay contact being some distance away is a concern. One issue is that the relay contact has to break a highish DC voltage and carries a DC current - that is pretty much the worst case application for a contact, and runs the risk of causing a voltage spike on the B+ feed to the OT's due to the primary inductances. Can I suggest relocating that contact in to either the bridge diode positive feed to the first capacitor, or in the PT HV winding feed to the diode bridge - as that allows the contact to see periodic zero current conditions, which helps a lot with arc quenching. That then should allow the final B+ filter cap to have a direct lead to the B+ feed of the OT.
Yes schem #6 is the step I consider worth making and ploting, as it routes the final bias cap C16 directly to the node used by the two cathode sense resistors and the B+ supply to the OT. That node is the appropriate common node for Vgk of each KT88, and the output stage B+.
Thanx for elaborating on the 650kHz issue. As 6A3sUMMER indicates, there may well be a problem with the B+ feed to the OT. Any OT primary signal current has to loop through the KT88's, the common cathode node, the last B+ filter cap, and back to the B+ feed on the OT. That loop contains the highest signal current level, and is imho, the most important loop to minimise for loop area and stray parasitic capacitance and inductance.
You indicate that common cathode node has a long lead run to the last B+ filter cap neg terminal, and that lead length can't really be reduced. Imho, the two cathode sense resistors need to separately run to the B+ cap negative, and that negative terminal should be a star for the bias 0V connection, and the 0V link to C2 and preamp circuitry, and the OT secondary winding ground. That star can have a link to chassis. The PE blocker circuit should go independently to chassis as it is effectively the PE connection to chassis, and has to handle any high current protective function.
The topic of the protective relay contact being some distance away is a concern. One issue is that the relay contact has to break a highish DC voltage and carries a DC current - that is pretty much the worst case application for a contact, and runs the risk of causing a voltage spike on the B+ feed to the OT's due to the primary inductances. Can I suggest relocating that contact in to either the bridge diode positive feed to the first capacitor, or in the PT HV winding feed to the diode bridge - as that allows the contact to see periodic zero current conditions, which helps a lot with arc quenching. That then should allow the final B+ filter cap to have a direct lead to the B+ feed of the OT.
Last edited:
The long cables are for sure a minus, but besides the 'Screen Grid Modifaction' link in post #59, I found quite some other schematics using a 1k resistor. Seems to be not the only amp with KT88 oscillation. This amp went into oscillation in the original configuration (no grid resistors on input and driver) by attaching an interconnect cable to the input and the scope told me 650 kHz....
Usually there is abaout 45 mA at the 10R resistors. I did the measurement without B+ on the left channel. Inputs shortened, plate-cathode 465 V, cathode 10R: 52 mA, sreen 1k: 4.5 mA
total dissipation: 465 V * 52 mA = 24.18 W
screen dissipation: (465 - 4.5 V) * 4.5 mA = 2.1 W
plate dissipation: 24.18 W - 2.1 W = 22.08 W
Is that correct ?
Actally it is quite simple. I bought an used AQ-1001 which was modified by the former owner and had burnt diodes. I am trying to learn as much as possible from you guys - THANKS A LOT FOR HELPING ME.
While learning I try to improve the amp whereever possible. It would be nice to keep the original chassis, but I realise that is a compromise.
If I can do anything different to make it easier for you, than please explain how, I will try to do it.
Yes. I was wondering if I could do something between ordinary trim-pot, transistor and auto-bias. Still wondering if I should go for auto-bias, transistor or trim-pot ?
I am very grateful for your extraordinary support, therefore the minimum I can do are these little things.
I fully agree, but the sim told me we are talking µA along a 12 cm fat cable.
The advantage of the current setup: I can go with a twisted pair from the bias supply to the last RC (C16) right at the trim-pots, while your proposal splits it up, but I will try.
Fully agree, see relay paragraph...
The cathode sense resistors are done that way, the bias thing I explained above, the C2 and preamp circuitry are done that way and the diodes to PE are done that way (see attached corrected schematic 5 from post #59). The diode bridge is probably misleading, sorry, I try to explain: the GND symbol is in real life a screw from the transformer mounting on the chassis. The secondary, the last HV cap (C14) and the diode bridge to PE are attached to this screw.
Why should the secondary winding go to the C14 (there is no NFB loop) and not to the chassis (I think that is safer), please explain ?
The relay contacts are not on my high priority list, first of all it should almost never happen, but Murphy says it will 😀. Second reason a relay is cheaper than the tubes, but burnt OPT is more expensive than tubes, therefore I need a better solution.
An easy way would be to elongate the HV AC to the relay and back to the diodes (see picture in the zip), but that way I have the switching noise close to the inputs, another disadvantage is that the relay cuts the AC, but the full charge of the caps keeps on burning the tubes for maybe 30 seconds longer than the current implementation.
Do I have to relocate the protection PCB and the bias PCB or do you have a better suggestion ?
Attachments
Any suggestions for positioning the bias supply PCB, the protection PCB and the choke for minimal hum/noise ?
Attached are two pictures for a possible mediocre solution.
Attached are two pictures for a possible mediocre solution.
From post #63:
Simulations are fine for some aspects, but it is difficult to make an accurate simulation for many performance aspects - this is a very detailed topic and typically requires a lot of verification testing and modelling. So imho, I suggest keeping any assessment of what a lead length may or may not do to just aim to keep the basic wiring concept as valid as possible - ie. take the positive end of C16 and the bias trim circuitry to the star node used by the output stage cathodes (and let all the bias circuitry and bias supply make the single link to 0V via that final bias connection path).
The OT secondary winding has an internal capacitance back to the OT primary winding - that capacitance can form a loop for higher frequency currents, and hence it is best to take to the primary side circuitry 0V. If there was a global negative feedback circuit (eg. to the PI or an earlier stage) then that secondary winding 0V would be a direct wire to the 0V of the primary side feedback circuit.
Your schem shows C14 going to the output stage 0V star, but your discussion says it actually goes to chassis??
The relay contact is imho a priority if it causes the OT B+ lead to be quite distant from C14 positive terminal. That relay should really be located near the PT HT winding leads, so that the contact is in series with one of the HV winding leads, along with a series fuse (not sure if you have a fuse in that location, but I would strongly suggest you add one). The energy in the B+ filter caps is fairly minuscule compared with many cycles of continued mains supply, and a PT secondary or OT primary is likely to survive a fault in the system in the time a well chosen secondary side fuse acts, and if a protection circuit can open the relay contact in a shorter time then even better. The B+ caps typically discharge in less than a second when the mains switch is turned off and the output stage continues to conduct idle current, due to hot cathodes - the fall rate of B+ would be even quicker if there was an over-current fault and the relay contact opened the HV secondary, or a fuse did.
If you haven't used a fuse in the PT HV secondary, or just took a guess at the fuse rating, then the link provides some additional design info on how to choose and select an appropriate fuse:
https://www.dalmura.com.au/static/Valve%20amp%20fusing.pdf
Simulations are fine for some aspects, but it is difficult to make an accurate simulation for many performance aspects - this is a very detailed topic and typically requires a lot of verification testing and modelling. So imho, I suggest keeping any assessment of what a lead length may or may not do to just aim to keep the basic wiring concept as valid as possible - ie. take the positive end of C16 and the bias trim circuitry to the star node used by the output stage cathodes (and let all the bias circuitry and bias supply make the single link to 0V via that final bias connection path).
The OT secondary winding has an internal capacitance back to the OT primary winding - that capacitance can form a loop for higher frequency currents, and hence it is best to take to the primary side circuitry 0V. If there was a global negative feedback circuit (eg. to the PI or an earlier stage) then that secondary winding 0V would be a direct wire to the 0V of the primary side feedback circuit.
Your schem shows C14 going to the output stage 0V star, but your discussion says it actually goes to chassis??
The relay contact is imho a priority if it causes the OT B+ lead to be quite distant from C14 positive terminal. That relay should really be located near the PT HT winding leads, so that the contact is in series with one of the HV winding leads, along with a series fuse (not sure if you have a fuse in that location, but I would strongly suggest you add one). The energy in the B+ filter caps is fairly minuscule compared with many cycles of continued mains supply, and a PT secondary or OT primary is likely to survive a fault in the system in the time a well chosen secondary side fuse acts, and if a protection circuit can open the relay contact in a shorter time then even better. The B+ caps typically discharge in less than a second when the mains switch is turned off and the output stage continues to conduct idle current, due to hot cathodes - the fall rate of B+ would be even quicker if there was an over-current fault and the relay contact opened the HV secondary, or a fuse did.
If you haven't used a fuse in the PT HV secondary, or just took a guess at the fuse rating, then the link provides some additional design info on how to choose and select an appropriate fuse:
https://www.dalmura.com.au/static/Valve%20amp%20fusing.pdf
Last edited:
From post #64:
Hooley dooley, that is quite some concentration of wiring!
Commenting may be a bit frivolous based on a quick view, but the following may give you something to mull over.
The choke may or may not be needed, depending on how well any hum can be minimised from a better understanding of how and why it is ingressing. In that location, the choke is likely to add more uncertainty than certainty as to what it can achieve. Removing the choke may allow better routing and separation of wiring, that may avoid accidental coupling of hum and signal feedback.
The OT wire loom seems very close to the input terminals and wiring - even with a local input selector switch. That area may well benefit from a steel shield barrier around the selector switch and input terminals.
The version B photo of the pcb shows two wires routed to a KT88 anode - one wire goes to the OT - does the other wire go to the local feedback? In general, all four KT88 anode wires need to be re-routed to be as far from anything else as possible, and certainly not mingled - if appropriate thay can route with the B+ lead from the OT.
One advantage of triode mode is that screen tap wiring doesn't have to be routed to the pcb, and those screen tap leads can be coiled and kept as distant from anything as possible (even preferably located back in a bell-end, or unsoldered if there are solder pads in the OT under the bell-end). The screen stopper can then be just local to the tube pin pads on the pcb.
The photos indicate the coupling cap pad is very close to the anode pad and trace on the pcb for one KT88 side - is that the situation?
The photo shows a fixed humdinger. Is there only one KT88 heater winding for both channels (ie. 4x KT88)? A humdinger pot is unlikely to tune out any residual hum that couples to a KT88 grid, but perhaps that is worthwhile confirming at some stage.
There is a long PE lead shown in the photos. Can you show how the mains wiring is routed with a photo? Any mains PE lead should go as directly as possible, and without undue length, to its own chassis connection. In your case you are aiming to insert a diode ground breaker in that PE link to chassis. It's unclear if you are just using a bridge of diodes, or other ancillary parts, and if you have based your approach on some particular ground breaker design.
Hooley dooley, that is quite some concentration of wiring!
Commenting may be a bit frivolous based on a quick view, but the following may give you something to mull over.
The choke may or may not be needed, depending on how well any hum can be minimised from a better understanding of how and why it is ingressing. In that location, the choke is likely to add more uncertainty than certainty as to what it can achieve. Removing the choke may allow better routing and separation of wiring, that may avoid accidental coupling of hum and signal feedback.
The OT wire loom seems very close to the input terminals and wiring - even with a local input selector switch. That area may well benefit from a steel shield barrier around the selector switch and input terminals.
The version B photo of the pcb shows two wires routed to a KT88 anode - one wire goes to the OT - does the other wire go to the local feedback? In general, all four KT88 anode wires need to be re-routed to be as far from anything else as possible, and certainly not mingled - if appropriate thay can route with the B+ lead from the OT.
One advantage of triode mode is that screen tap wiring doesn't have to be routed to the pcb, and those screen tap leads can be coiled and kept as distant from anything as possible (even preferably located back in a bell-end, or unsoldered if there are solder pads in the OT under the bell-end). The screen stopper can then be just local to the tube pin pads on the pcb.
The photos indicate the coupling cap pad is very close to the anode pad and trace on the pcb for one KT88 side - is that the situation?
The photo shows a fixed humdinger. Is there only one KT88 heater winding for both channels (ie. 4x KT88)? A humdinger pot is unlikely to tune out any residual hum that couples to a KT88 grid, but perhaps that is worthwhile confirming at some stage.
There is a long PE lead shown in the photos. Can you show how the mains wiring is routed with a photo? Any mains PE lead should go as directly as possible, and without undue length, to its own chassis connection. In your case you are aiming to insert a diode ground breaker in that PE link to chassis. It's unclear if you are just using a bridge of diodes, or other ancillary parts, and if you have based your approach on some particular ground breaker design.
Last edited:
Will do, thank you.
Sorry for my bad English. The schem is correct, the minus terminal of C14 is the 0 V. From the C14 minus terminal (the 0 V) runs a wire(actually 2 to have less resistance) to the screw that makes the chassis connection.
Thanks alot for the link !!! Ofcourse I will read the other articles as well, hopefully my old brain can remember a lot 🙂.
Please have a look at Fig. 1, column 2.1x, I think those values should be swapped.
I have no secondary fuse in the moment (will add it and a NTC), there is a M 5A primary fuse, the amp is drawing a little bit less than 0.7 A.
The original amp had 1N4007 for the HV, as you probably noticed, the PCB is brown, therefore I replaced them with UF5408. If I understood you correct, I should replace the UF5408 with 2*UF4007 ?
The cathode fusing with the parallel Zener sounds very interesting to me. The protection circuit compares the voltage across the 10R cathode sensing resistors to a series string of 3 1N4007 and triggers at 2V4 (240 mA). As you pointed out to me earlier the relay causes many problems, replacing it by fuses and Zeners would free a lot of space. Do you suggest removing the protection relay PCB and replace it by the fuses ?
Now you understand why I was shy to show pictures 😀.
Yesterday I did already some testing, tried all possible positions of the choke. All tubes removed, no B+ connection. Only the C13+33R+C14 are powered. the leads of the choke connected to the DMM. Sitting on the chassis, the minimum value I get is 8 mVrms, lifting the choke 6 cm away from the chassis I get 1.8 mVrms and without mains I read 1.3 mVrms. I guess that means you are right again 😀 and I can forget about the choke.
No the local feedback 2M2+10nF was temporary soldered to the socket pins and is history. Assuming you are talking about the orange and violet wires, one wire from the OPT, the other wire from the T/UL switch situated between the two big L+R PCBs. Grey and white are the corresponding screen wires.
Did that yesterday 😀. Should I stay with the 1k or try something like constant voltage ?
Yes, the 'orange' anode is very close to the grid cap - see attachment 'grid cap-anode' showing proposed changes.
Original the GND is one pcb trace running all over the place and branched wherever convenient. I cutted this GND trace many times to realise the returns as shown in the schematic.
There are two windings for L+R (2x KT88 + input and driver tube). I tried humdinger pot earlier and I could tune it worse than the fixed solution, but not better, but that was before I realised your suggested improvements. I tried DC heating two times before and had no change in hum/noise compared to AC at that time.
I attached the photo again, if you want a better one, please tell me. Bottom right is mains inlet, red/yellow wire runs to mains switch (bottom left) and back to PT. Close to the mains inlet the yellow changes to brown (red/brown) to PT. I added the yellow to have some canceling on the way to the mains switch, as original there was only the red wire running to the switch. The primary fuse is just above the mains inlet situated.
After reading this : The dozens schemes to wire an amp... and this: The dozens schemes to wire an amp... I disconnected the link from PE to mains inlet screw and used the bridge to connect 0 V and PE.
I use only diode bridge as shown in the schematic. I tried a cap for RF and didn't notice any difference (if you suggest I will add it). What I don't understand is the usual 10R. I mean, I want to stop the ground loop current by using the diode bridge, why should I open a path by adding the 10R ?
I'm not sure what I said that made you think that I suggested changing the UF5408?
Given your present protection circuit aims to sense cathode current, then it would need to adequately determine a fault condition, and not trip due to a current peak that could occur in normal operation and that wasn't damaging to the valve or transformers. You could well have a great trip mechanism, but it could still be unwieldy as you have to locate it, and power it, and that ends up requiring many parts and additional cabling that could be a risk.
The alternate option you raise is to continue to use a cathode sense resistor, and then add in a series fuse/zener network. That would be required for each cathode, and may be practical to insert, although maybe only if you can purchase fuses with pig-tails, and of a value/style that gave confidence of blowing in a coordinated manner.
Wrt the UL/T switch, that may seem a simple switch to include, but until you have a well performing amp, imho it adds an uncertainty due to superfluous wiring that could cause inadvertent coupling.
The photo showing grid coupling cap and anode pad changes, seems appropriate to lessen the chance of some interaction.
It is not normally allowable to connect a PE cable to chassis with a fixed bolt that is used by other chassis functional connections. Is there another wire soldered to the diode blocker bridge?
The photo shows all 4 cathode 0V leads going to a common star that is not the common C14 capacitor negative terminal. The C14 negative terminal may seem like an unwieldy location to route to, but at least you can use ring lugs.
Has the active wiring to the mains switch been twisted with the neutral along its route? It can be difficult to screen a long active wire like that. Another technique may be to run just the active in the chassis bend, and restrain it to the bends, such that it is as far from any other part or wire as practical. Although twisting can work, if that means that parts of the active wire are significantly closer to other parts then some doubt arises - but at this stage any such wire layout is likely a negligible contributor if testing is just on the other channel.
Another photo of the total underneath would be good to see if you were to simplify the circuit to triode mode, and remove the UL/triode switch wiring and the OT screen tap wiring.
Last edited:
A clever guy wrote in https://dalmura.com.au/static/Hum%20article.pdf:
"An advantage of using series ss diodes (for PIV
reasons) also effectively halves the stray coupling
capacitance. Using a diode of lower current rating
also helps – a 1A rated 1N4007 or UF4007 has
capacitance (15pF @ 4Vr) about 10 times less than
a 6A rated diode (P600M, 150pF @ 4Vr)."
Which in my case would be UF4007 17pF/2=8.5pF compared to 36pF for the UF5408. Therefore I thought I should try it. Wrong idea ?
I need to keep the cathode sense resistor to adjust the bias. I can use one of the following three options to protect the tubes:
1) the protection circuit: advantage - precise triggering at 240 mA, disadvantage - needs a lot of space and additional wiring
2) the fuse + Zener: advantage - easy to implement using pig tail fuse holder, disadvantage - I am not sure if the fuse saves the tube or vice versa.
3) low wattage cathode resistor: advantage - needs no space, disadvantage - biggest uncertainty, whether it blows in time or not.
If you don't tell me otherwise, I tend to favor the protection circuit, modified as per your suggestion to break the HV AC and I would add a fuse in the HV AC leg to protect the HV secondary. Do you agree ?
It is actually a very flimsy connection, violating the rules, but it has a China Export sticker on it...😡
The black wire is the +/- connection to make the diode bridge work, the yellow wires are in/out, nothing else connected.
The photo was taken when there was the old connection (star near input - school) in place, it is all gone in the meantime.
I might be able to add an additional RC for the L and the R channel, that is from C14 via R(L) to C(L) and via R(R) to C(R). This would allow shorter wires from the left pcb(L) to the C(L) and from the right pcb(R) to the C(R) and give a little bit better channel separation. Do you like that ?
There is no defined active or neutral, as the mains plug can be inserted either way. There should be a double pole switch, but there is only a single. I added the yellow wire to have active/neutral cancellation, but at that time there was more hum/noise and I could not see any benefit on the DMM.
Measuring the amp without any secondary connected (no HV, no bias, no tubes,...) there is a big difference between amp off/on in the left channel OPT, but nothing changes if I move the red/yellow(brown) primary wires.
Done already, photo coming soon.
In the moment I am evaluating a flux band - seems to help.
Using any particular diode requires a recognition of its operating conditions compared to its specs. UF4007 is certainly fine for many valve amps, but caution is needed when the power requirement increases, and that may well be a concern for KT88 type valves, and for quads/stereo. Many would say just stick with the UF5408 as it is unlikely that those diodes would be stressed in your amp, and it is unlikely that you would be able to measure any comparative difference (with respect to rectification noise). If you were keen to do a comparison, or just make a swap for your own reasons, then you could set up a PSUD2 assessment, but note that assessment issues can get a bit technical, as indicated by the discussion in link on how much current a UF4007 is likely to handle.
https://www.dalmura.com.au/static/Power%20supply%20issues%20for%20tube%20amps.pdf
Wrt to cathode over-current protection, 'a bird in the hand is worth 2 in the bush' does seem reasonable, especially if you can route wires such that they aren't a noise concern. You also likely have the ability to tweak the trip setting fairly easily. An in-line fuse is definitely advisable as a backup, as contacts can stick as a failure mechanism.
Channel separation performance may well have multiple contributions, and yes having some channel B+ impedance that assists in constraining output stage signal currents to just the channel itself should help.
https://www.dalmura.com.au/static/Power%20supply%20issues%20for%20tube%20amps.pdf
Wrt to cathode over-current protection, 'a bird in the hand is worth 2 in the bush' does seem reasonable, especially if you can route wires such that they aren't a noise concern. You also likely have the ability to tweak the trip setting fairly easily. An in-line fuse is definitely advisable as a backup, as contacts can stick as a failure mechanism.
Channel separation performance may well have multiple contributions, and yes having some channel B+ impedance that assists in constraining output stage signal currents to just the channel itself should help.
I'm not sure I appreciate what that means - can you elaborate a bit please.
I am tryingto figure out how big the induction of the PT into the OPTs is. I disconnected all the secondary windings from the pcbs and left them open circuit. Any voltage on the primary or secondary of the OPT would show the influence of the PT and the effectiveness of a flux band on the PT and/or OPT. Both channels clearly show that a lot of hum/noise is coming from the PT.
I also wanted to see the influence of the mains cable (red/yellow[brown]). I tried to move the cable as far as possible away from the pcb and I couldn't measure any difference. My conclusion was the induction of the PT is much stronger, therefore I cannot see a benefit of moving the mains cable further away from the OPT cables.
I am lost...
I disconnected all the secondary windings from the pcbs and left them open circuit. Any voltage on the primary or secondary of the OPT should show the influence of the PT. Both channels clearly show that a lot of hum/noise is coming from the PT. I added a flux band on the PT which seems to help and on the OPTs with less improvement, but still a little bit better again.
The DMM measures 0.073 mVrms on both channels 8R secondary without any load and no power. When I power the amp(=the primary of the PT), no secondary is connected(=no B+, no Bias) the left channel has 0.580 mVrms, the right channel 0.215 mVrms.
Left channel center tab of the OPT to anode wire 7 mVrms.
Right channel center tab of the OPT to anode wire 3 mVrms.
I unsoldered the secondary winding wires from the speaker terminals on the left side and moved them around to see if there is any change in the readings - I couldn't dedect any difference, same result with moving the mains wiring close to the secondary winding or further away.
Both OPT are 90° to the PT. Drilling the red(center tab) and the orange+violet(anode) wires together seems to be a tiny little bit better than just running the wires parallel. Connecting the flux bands of the OPTs to the PE side of the diode bridge and the flux band of the PT to the 0 V side of the diode bridge made a tiny improvement.
I have no idea why the left side is so much worse than the right side....please help.
I disconnected all the secondary windings from the pcbs and left them open circuit. Any voltage on the primary or secondary of the OPT should show the influence of the PT. Both channels clearly show that a lot of hum/noise is coming from the PT. I added a flux band on the PT which seems to help and on the OPTs with less improvement, but still a little bit better again.
The DMM measures 0.073 mVrms on both channels 8R secondary without any load and no power. When I power the amp(=the primary of the PT), no secondary is connected(=no B+, no Bias) the left channel has 0.580 mVrms, the right channel 0.215 mVrms.
Left channel center tab of the OPT to anode wire 7 mVrms.
Right channel center tab of the OPT to anode wire 3 mVrms.
I unsoldered the secondary winding wires from the speaker terminals on the left side and moved them around to see if there is any change in the readings - I couldn't dedect any difference, same result with moving the mains wiring close to the secondary winding or further away.
Both OPT are 90° to the PT. Drilling the red(center tab) and the orange+violet(anode) wires together seems to be a tiny little bit better than just running the wires parallel. Connecting the flux bands of the OPTs to the PE side of the diode bridge and the flux band of the PT to the 0 V side of the diode bridge made a tiny improvement.
I have no idea why the left side is so much worse than the right side....please help.
The flux band is soldered and has a wire to be attached to 0 V or PE.
With the amp bottom side up, mains on primary, no secondary voltages I get about 2.5 mVrms between right channel anode (clip on center and one anode, other anode nothing attached). Left channel 10 mVrms
.
Loosening the nuts of the mounting bolts of the OPT (but there is still the weight of the OPT) I get 2.8 mVrms. Do I need rubber grommets ?
It seems I found a new way to measure the tension in a screw 🙂. The higher the tension on the pot mounting screws, the smaller the induced voltage. There is about 20% difference between tight and super tight in this case. Which is just enough to get equal channel readings...should I buy serrated lock washers for the pot screws for better contact or do you think it is all about the microscopic(?) gap between pot and chassis ?
With the amp bottom side up, mains on primary, no secondary voltages I get about 2.5 mVrms between right channel anode (clip on center and one anode, other anode nothing attached). Left channel 10 mVrms
. Loosening the nuts of the mounting bolts of the OPT (but there is still the weight of the OPT) I get 2.8 mVrms. Do I need rubber grommets ?
It seems I found a new way to measure the tension in a screw 🙂. The higher the tension on the pot mounting screws, the smaller the induced voltage. There is about 20% difference between tight and super tight in this case. Which is just enough to get equal channel readings...should I buy serrated lock washers for the pot screws for better contact or do you think it is all about the microscopic(?) gap between pot and chassis ?
Sorry...see attachments.
Would you be able to post a photo of the chassis top, but looking from the side, so as to get a better appreciation of how the transformer cores are mounted to the chassis, and how each 'pot' covers each transformer and wires.
The core and clamping bolts seem to be varnish dipped. Can you clarify if each clamping bolt has a fibre washer between the outer-most core lamination or cross piece, and the bolt's washer/nut, and at each end of the clamping bolt ? The mounting bolts don't appear to be insulated.
I would add a safety caution about how you have applied the flux band. Typically a flux band width is made about 1/3 the width of the winding. One benefit of the external band width not extending to the winding edge is that there is no chance of reduced creepage between any winding layer end turns and the added band.
The core and clamping bolts seem to be varnish dipped. Can you clarify if each clamping bolt has a fibre washer between the outer-most core lamination or cross piece, and the bolt's washer/nut, and at each end of the clamping bolt ? The mounting bolts don't appear to be insulated.
I would add a safety caution about how you have applied the flux band. Typically a flux band width is made about 1/3 the width of the winding. One benefit of the external band width not extending to the winding edge is that there is no chance of reduced creepage between any winding layer end turns and the added band.
See attachments, if you need more please tell me (I didn't really understand what you are asking for regarding the 'pot'). The 'pot' photo should show that the black color on the aluminum is covering the contact area to the stainless steel chassis, but the threads of the screws cut into the aluminum (no color).
There is only metal, no fibre washers nowhere, nothing is insulated. Bolts, nuts and washers are ferro magnetic.
Didn't know that, it was already finished when I read your article.
I don't know if there is a need for a flux band or if a 'pot' is good enough ?
Should I either remove it or should I make a 1/3 width version ?
Thanks for the photos - that clarifies the mounting arrangement of the transformer, and also the 'pot' (which originally had me a bit uncertain as to what you were referring to).
An additional comment is that there may be circulating currents via the mounting bolts, as they pass through the core effectively as a shorted turn. Normally, bolts that pass through holes in a core are insulated, usually at the washer/nut ends, and sometimes also using a sleeve over the bolt. That may influence the core loss a bit, and may increase the self inductance (ie. radiated field).
You may want to assess whether the end nuts can be taken off, but that may be impractical if the varnish is impregnated in to all the voids and threads.
Another comment is that there may be some room on the inside of the pot to fit steel plates, as a means to reduce stray field coupling. The pot itself will effectively do nothing to constrain stray magnetic flux (which is the coupling mechanism between windings on separate cores), as it is aluminium.
Do you have a photo of the core and winding before the flux band was applied? If you kept the copper covering as is, I can suggest that one way to reduce the risk that there could be leakage or breakdown to the flux band is to do an insulation resistance check (but that type of test is also risky). My concern is not the width of the flux band.
An additional comment is that there may be circulating currents via the mounting bolts, as they pass through the core effectively as a shorted turn. Normally, bolts that pass through holes in a core are insulated, usually at the washer/nut ends, and sometimes also using a sleeve over the bolt. That may influence the core loss a bit, and may increase the self inductance (ie. radiated field).
You may want to assess whether the end nuts can be taken off, but that may be impractical if the varnish is impregnated in to all the voids and threads.
Another comment is that there may be some room on the inside of the pot to fit steel plates, as a means to reduce stray field coupling. The pot itself will effectively do nothing to constrain stray magnetic flux (which is the coupling mechanism between windings on separate cores), as it is aluminium.
Do you have a photo of the core and winding before the flux band was applied? If you kept the copper covering as is, I can suggest that one way to reduce the risk that there could be leakage or breakdown to the flux band is to do an insulation resistance check (but that type of test is also risky). My concern is not the width of the flux band.
Sorry for my bad English. I know the expression 'potted transformer' and the dictionary confirmed 'pot'. What is the correct word for it ?
I can not loosen the EI-core nuts, I could cut the nuts and make everything new, but I am unsure if this is worth it. I could try to enlarge the chassis holes and use rubber grommets for the chassis mounting.
I don't understand the big difference between L and R channel and why loosening the nuts has such a dramatic effect although there is still electrical contact ?
The edge/corner of the EI slides into the corner recess. I could only add 3 (maybe 4) mm thick plates on the 4 sides, nothing on top or bottom. Is this worth trying ?
The flux band is for sure an improvement compared to the naked OPT/PT, but there is a considerable reduction of the induced voltage when I put the 'pot' on the OPT and/or PT in spite of the flux band. This makes me question if a flux band is of any use if there is a 'pot' ?
Sorry, no. Looked like paper(?) insulation, everything is covered with black varnish(?).
O.k. I understood that if the bobbin has a width of 4.5 cm, the flux band width should be 4.5 / 3 = 1.5 cm.
Do you say if the winding wire has a diameter of 0.9 mm, than the thickness of the flux band should be 0.9 / 3 = 0.3 mm ? Wrong again ?
I am not really keen on removing the flux band on one OPT, but if you think it's worth it I will. Please advise and thank you again for helping me !
Any incremental improvement sounds like it is worth doing, if you can make the effort and can make a measurement that confirms the improvement.
I just checked on the flux band info I have (from a book by Nave) - he indicates 50% of the winding width is best, however that refers to ferrite cored transformers for switchmode power supplies. If you were to redo it, then perhaps go out to almost the edge of the winding width, but leave say 6mm margin on each side. You need to determine if it is safe or not.
The mounting bolts may be a contributor, by way of allowing a circulating current path via the ss chassis. Insulating the 4 mounting bolts of each transformer (especially the power transformer) to the amp chassis may be a worthwhile test if you can practically achieve that, as that may also indicate why there is a measureable difference between what is coupled to the two OT transformers.
I just checked on the flux band info I have (from a book by Nave) - he indicates 50% of the winding width is best, however that refers to ferrite cored transformers for switchmode power supplies. If you were to redo it, then perhaps go out to almost the edge of the winding width, but leave say 6mm margin on each side. You need to determine if it is safe or not.
The mounting bolts may be a contributor, by way of allowing a circulating current path via the ss chassis. Insulating the 4 mounting bolts of each transformer (especially the power transformer) to the amp chassis may be a worthwhile test if you can practically achieve that, as that may also indicate why there is a measureable difference between what is coupled to the two OT transformers.