Hello,
First, again thank you 😊
Grover, this is a good news I'll go back to AC
Mircea, a lot of GND as been added in case, you are totally right
trobbins, for the separation it was a test to kind of isolate more the pads but in the end i just put more space and removed the copper plane around.
About the KT88 pin outs, number 1 is back but I tested one of my JJ tube and nothing is connected to it. I changed V4 traces and put a lot more space between them, I also add two cut to put something to isolate from arcing.
R10/R9 (470K) are now closer and I think better placed
No more traces "into" the socket, all routed around it.
I removed the copper under C4/C3.
I really hope to have understand what you meant
First, again thank you 😊
Grover, this is a good news I'll go back to AC
Mircea, a lot of GND as been added in case, you are totally right
trobbins, for the separation it was a test to kind of isolate more the pads but in the end i just put more space and removed the copper plane around.
About the KT88 pin outs, number 1 is back but I tested one of my JJ tube and nothing is connected to it. I changed V4 traces and put a lot more space between them, I also add two cut to put something to isolate from arcing.
Here you lost me 😆 but, I tried to put them close to V2.
R10/R9 (470K) are now closer and I think better placed
No more traces "into" the socket, all routed around it.
I removed the copper under C4/C3.
I really hope to have understand what you meant
That looks better to my eyes. My only continuing concern relates to C5, C6 that have a gnd plane under them, and so would have more parasitic capacitance to that gnd than say a vintage amp using point to point wiring. Similar to what you have done for C3, C4, you could remove the copper under C5, C6, as you have a reasonable amount of copper joining the various gnd plane regions. If you were concerned about not having as good a general ground plane then you could add thick traces to the other side of the board where you think it is marginal (eg. gnd ends of R7 and R10).
I think it is still in your interest to do a free-hand sketch of a gnd connection schematic - as a double check to know what gnds connect to where for all the parts etc off the pcbs.
I think it is still in your interest to do a free-hand sketch of a gnd connection schematic - as a double check to know what gnds connect to where for all the parts etc off the pcbs.
Hello,
Well I think this time the PCB is done, I removed the copper under C5/C6 and around the heaters AC pads / traces, I also used some VIA around V1/V2, more more ground and some other placement modifications.
I'll follow your advice about a free-hand sketch, with the added gnd at bottom it should be fine.
If all good I'll order it 😊
Well I think this time the PCB is done, I removed the copper under C5/C6 and around the heaters AC pads / traces, I also used some VIA around V1/V2, more more ground and some other placement modifications.
I'll follow your advice about a free-hand sketch, with the added gnd at bottom it should be fine.
If all good I'll order it 😊
A comment about heater wiring seems relevant. In general it is better to screen any such heater traces as much as practical when nearby to input stage circuitry - for a pcb that is typically done by having gnd plane nearby - both next to the traces on the trace side, and underneath if that is practical (which it is) - this is a situation where having gnd plane nearby and hence capacitance to gnd is a good thing (not a bad thing as per signal circuitry). Not shielding the heater traces means the heater voltage can more easily couple by stray capacitance into signal lines. There still needs to be adequate separation between heater and gnd, given that there is an option for the heater supply to be elevated.
If you have 'hot' parts then that is a situation where adding copper 'lands' can act as heatsinking to alleviate temperature rise of parts and pads and pcb in close proximity - often seen on old pcbs as a region of heat stressed brown pcb. Perhaps the parts I can easily see that could have moderate power dissipation would be the HT diodes, the HT rail dropper resistors, and some of the KT88 terminals. What was your intent with the copperless vias under the dropper resistors? WRT the HT diodes, the only practical change to suppress thermal resistance between parts and to ambient would be to use similarly thick traces on both side of the pcb (from entry pads through to F1 pad). Not all diode options have large bodies, and there are other benefits in using as 'small' a diode as practical. I think you are using added octal sockets, which will go some way to lowering the temp rise of pads/traces on the pcb, and I'm not sure how much of an issue this is for pcb based designs as I don't typically come across them.
I also just noticed that your bias supply doesn't have any bleed resistance on the pcb itself - that is only a risk if you power up the pcb without the bias circuitry.
If you have 'hot' parts then that is a situation where adding copper 'lands' can act as heatsinking to alleviate temperature rise of parts and pads and pcb in close proximity - often seen on old pcbs as a region of heat stressed brown pcb. Perhaps the parts I can easily see that could have moderate power dissipation would be the HT diodes, the HT rail dropper resistors, and some of the KT88 terminals. What was your intent with the copperless vias under the dropper resistors? WRT the HT diodes, the only practical change to suppress thermal resistance between parts and to ambient would be to use similarly thick traces on both side of the pcb (from entry pads through to F1 pad). Not all diode options have large bodies, and there are other benefits in using as 'small' a diode as practical. I think you are using added octal sockets, which will go some way to lowering the temp rise of pads/traces on the pcb, and I'm not sure how much of an issue this is for pcb based designs as I don't typically come across them.
I also just noticed that your bias supply doesn't have any bleed resistance on the pcb itself - that is only a risk if you power up the pcb without the bias circuitry.
Last edited:
Hello 😀
Oh I thought the magnetic field generated by the AC traces will pollute the mass GND, so this time it's the opposite, I putted back the copper and more distance between AC pads and HT, I also added traces under to double them, I reduced from 2mm to 1mmX2.
The copper less zone around the HT diode was with the same thought about a possible field due to the AC coming into the diodes, so for the question of dissipation I doubled the traces of all the diodes until F1.
Should I include this area into the GND copper plane, does it work as for the AC Heaters even if this is HT ? Space is 1.1mm so it should be okay but I can put more space if needed.
For the KT88 and also now 6SN7 I'm gonna use those:
So they are way heavier than the previous one that I wanted to use, concerning the traces around them, I reduced then a bit from 2mm to 1.5mm width but now they are doubled so 3mm it's gonna be more than ok I can go back to 2mm (4mm) if necessary
For the global strength of the PCB and dissipation I'm wondering if I shouldn't add an another Copper plane in the other face but not link to GND with few holes to help the heat to leave the board
About the bias bleed resistance, do you suggest to add a big like 220K 470K resistor link between the bias voltage and the gnd ?
And for the other bleed resistance every big filter cap will have 470K to keep then without any voltage left when the amplifier is off and for safety 😆
This is the last version of the PCB:
changelog:
-1 Ohm probes better placed and also the trace,
Future change MAYBE:
and shall be good 😊
Cheers
Alexis
Oh I thought the magnetic field generated by the AC traces will pollute the mass GND, so this time it's the opposite, I putted back the copper and more distance between AC pads and HT, I also added traces under to double them, I reduced from 2mm to 1mmX2.
The copper less zone around the HT diode was with the same thought about a possible field due to the AC coming into the diodes, so for the question of dissipation I doubled the traces of all the diodes until F1.
Should I include this area into the GND copper plane, does it work as for the AC Heaters even if this is HT ? Space is 1.1mm so it should be okay but I can put more space if needed.
For the KT88 and also now 6SN7 I'm gonna use those:
So they are way heavier than the previous one that I wanted to use, concerning the traces around them, I reduced then a bit from 2mm to 1.5mm width but now they are doubled so 3mm it's gonna be more than ok I can go back to 2mm (4mm) if necessary
For the global strength of the PCB and dissipation I'm wondering if I shouldn't add an another Copper plane in the other face but not link to GND with few holes to help the heat to leave the board
About the bias bleed resistance, do you suggest to add a big like 220K 470K resistor link between the bias voltage and the gnd ?
And for the other bleed resistance every big filter cap will have 470K to keep then without any voltage left when the amplifier is off and for safety 😆
This is the last version of the PCB:
changelog:
-1 Ohm probes better placed and also the trace,
- KT88 double traces,
- 6SN7 Heater double traces,
- HT diodes doubles traces,
- Amp PCB a bit wider +5mm to improve space for heaters V3
- 6SN7 Socket also changed
- VIA's removed and traces improved to limit the GND copper plane
- More ground again at the bottom,
Future change MAYBE:
- One more copper plane for GND + holes ?
- Pin 6 will be back for the KT88 for more support as traces are finer
and shall be good 😊
Cheers
Alexis
Last edited:
Orion24, you could bring the gnd plane closer to the KT88 heaters, however the KT88's have much higher AC voltages on the anode and screen, and separation distance is being used to alleviate those signals from coupling into preamp circuitry. So that is the practical balance being applied.
Those tube sockets do look good, including their mounting wings to minimise stress into the pcb pads. And yes, including a pad for pin 6 (and even with the heater trace through that pad) seems simple and effective, as the 7027 is the only valve it would concern afaik.
Your wired connections to the V1/2 heater pads should provide significant thermal conduction away from the tube socket pins, as the pcb links are short and wire is a good heatsink. Using 2oz copper, and traces on both sides should also be noticeable for the V3/4 traces removing heat to wiring. If you get concerned during testing about temperature rise of socket pads, then the backup plan would be to add a run of TCW from V3/4 socket pads to wiring pads - for some applications that require very low resistance, or lower thermal resistance, manufacturers may lay more solder on to traces/planes, and certainly diy efforts have laid TCW on top of traces even though it may not be aesthetically pleasing.
BTW, what does this mean " I reduced from 2mm to 1mmX2." ?
Those tube sockets do look good, including their mounting wings to minimise stress into the pcb pads. And yes, including a pad for pin 6 (and even with the heater trace through that pad) seems simple and effective, as the 7027 is the only valve it would concern afaik.
Your wired connections to the V1/2 heater pads should provide significant thermal conduction away from the tube socket pins, as the pcb links are short and wire is a good heatsink. Using 2oz copper, and traces on both sides should also be noticeable for the V3/4 traces removing heat to wiring. If you get concerned during testing about temperature rise of socket pads, then the backup plan would be to add a run of TCW from V3/4 socket pads to wiring pads - for some applications that require very low resistance, or lower thermal resistance, manufacturers may lay more solder on to traces/planes, and certainly diy efforts have laid TCW on top of traces even though it may not be aesthetically pleasing.
BTW, what does this mean " I reduced from 2mm to 1mmX2." ?
Another thread highlighted the issue of trimming distortion once a pcb is populated and amp is being tested. The typical differences in PI and driver triode and KT88 gains will I suggest inevitably lead to imbalance that can be minimised with appropriate tweaking of certain resistor values. The original Williamson utilised a pot in the driver B+ feed. There are other resistors that can similarly be used for gain trimming.
The pcb already has some ability to trim, by for example clipping a pot across say R12 or R13, and adjusting for minimum output harmonics, and then fitting a 'Select on Test' part (made easier if pads were available, or posts were available for clipping to and then soldering to).
The pcb already has some ability to trim, by for example clipping a pot across say R12 or R13, and adjusting for minimum output harmonics, and then fitting a 'Select on Test' part (made easier if pads were available, or posts were available for clipping to and then soldering to).
Thank you all 😊
Trobbins, I moved closer the GND plan to the KT88 but I took care about the HT pins with "a lot" of distance especially with pins 3/4, and now 2/6 are rooted together to have copper around the pad.
I think adding solder on some traces will not be necessary as they are width enough 2mm (HT and Heaters) plus they are doubled and as the copper will be 2OZ.
For the possible imbalance I added 2 another Trim-pot of 10K above R12/R13 and changed them for 44K to have a range of 44K - 54K is it enough ? And for the power of those TRIM I can only get 500mW usually I like to use large tolerances but 1W are not findable for 10K and the max is gonna be 250mW.
This is the last release:
And caps are arrived 😀
Trobbins, I moved closer the GND plan to the KT88 but I took care about the HT pins with "a lot" of distance especially with pins 3/4, and now 2/6 are rooted together to have copper around the pad.
I think adding solder on some traces will not be necessary as they are width enough 2mm (HT and Heaters) plus they are doubled and as the copper will be 2OZ.
It was just to say I doubled the 6SN7 preamp heater traces but I putted back 2mm doubled just for more dissipation.
For the possible imbalance I added 2 another Trim-pot of 10K above R12/R13 and changed them for 44K to have a range of 44K - 54K is it enough ? And for the power of those TRIM I can only get 500mW usually I like to use large tolerances but 1W are not findable for 10K and the max is gonna be 250mW.
This is the last release:
And caps are arrived 😀
Tim, I can't imagine how dumb you must think I'm 😆 Now I see what you meant, I just find the original schematic...
So I changed the schematic and the PCB to be right
Now I have to wait the sockets before launch the PCB to check if traces are good
Have a nice day
Alexis
So I changed the schematic and the PCB to be right
Now I have to wait the sockets before launch the PCB to check if traces are good
Have a nice day
Alexis
Hello
There is one thing I have observed in time with this kind of schematic which I think it’s worth mentioning.
While adjusting the trimmers R24/R25 for less distosion you might affect the harmonic behaviour of that distorsion.
What I am saying is that if you only aim for minimum THD you might end up with a dominant H3 which is not what you really want from a tube amplifier.
Anyway, I will follow this thread and we can discuss that after you populate the PCB’s.
Cheers,
Mircea
There is one thing I have observed in time with this kind of schematic which I think it’s worth mentioning.
While adjusting the trimmers R24/R25 for less distosion you might affect the harmonic behaviour of that distorsion.
What I am saying is that if you only aim for minimum THD you might end up with a dominant H3 which is not what you really want from a tube amplifier.
Anyway, I will follow this thread and we can discuss that after you populate the PCB’s.
Cheers,
Mircea
For casing look around for pretty old? or newer gear that you like. Who says you can't build your amp into a toaster or tabletop oven?
Cheers!
Cheers!
Good question !
I've build a PC in a wine crate and assembly an class D amp in inverter welder machine because...why not ? (or wine hot )
I totally agree with this, as long as THD is below 3%, I will actually optimize to maximize H2.
Yes, I said “maximize” and I realize this will make some of you cringe. But, hey, personal preference is not Engineering nor Science.
I have built quite a few amps in 19” server enclosures, cheap and well ventilated. Not saying it looks great or that it is convenient, though.
I dream of building an amp in an enclosure that would look like a cask, a locomotive or the body of a musical instrument.
Humm this is a good question, for sure as I also make my guitars I'll put some wood ! I took a big Hammond chassis, it need to be a bit re-enforced and why not add a faceplate with Brushed Aluminum and wood on both side to act as foot but yeah everything is good to became a chassis 😆
The pictures I sent you showed a large Hammond chassis, entirely made steel. Probably 10 pounds on its own.
Working with steel is very different than working with aluminum. I struggled!
But the thing is a tank and it could go through walls without a scratch. (I should remove the tubes before trying that. 🤣)
So, if you add wood sidings to it and other trim decorations, it’ll be both beautiful and strong!
Before connecting and applying global negative feedback, then . . .
1. R25 can be adjusted for equal plate voltage.
(Un-matched triodes that tie both cathodes to a common 390 Ohm self bias resistor, will have un-equal DC plate voltages).
But adjusting R25 also effectively changes the 2nd harmonic distortion of the output stage.
R24 plus R8 total Ohms, versus R7 Ohms adjusts the relative amplitudes of the concertina phase splitter output, which makes the 2nd harmonic distortion of the later stages a variable.
With two pots: R24 and R25, they both cause variable 2nd harmonic distortion at the output (interaction).
2. Are you designing the amplifier to have 2nd harmonic distortion, and then applying global negative feedback:
That will eliminate most of the 2nd harmonic distortion that you intended to design into the amplifier.
Seems like a paradox.
3. I you want a simple way to get 2nd harmonic distortion, you might try designing single ended amplifiers.
Just be sure that the 2nd harmonic distortion of the input/driver stage is not the same as the 2nd harmonic distortion of the output stage . . .
or else l the 2nd harmonic distortion will be cancelled.
4. My balanced amplifier has no global negative feedback, and the 2nd harmonic distortion and 3rd harmonic distortion are both 0.3% or less.
Their percentages are almost equal to each other.
1. R25 can be adjusted for equal plate voltage.
(Un-matched triodes that tie both cathodes to a common 390 Ohm self bias resistor, will have un-equal DC plate voltages).
But adjusting R25 also effectively changes the 2nd harmonic distortion of the output stage.
R24 plus R8 total Ohms, versus R7 Ohms adjusts the relative amplitudes of the concertina phase splitter output, which makes the 2nd harmonic distortion of the later stages a variable.
With two pots: R24 and R25, they both cause variable 2nd harmonic distortion at the output (interaction).
2. Are you designing the amplifier to have 2nd harmonic distortion, and then applying global negative feedback:
That will eliminate most of the 2nd harmonic distortion that you intended to design into the amplifier.
Seems like a paradox.
3. I you want a simple way to get 2nd harmonic distortion, you might try designing single ended amplifiers.
Just be sure that the 2nd harmonic distortion of the input/driver stage is not the same as the 2nd harmonic distortion of the output stage . . .
or else l the 2nd harmonic distortion will be cancelled.
4. My balanced amplifier has no global negative feedback, and the 2nd harmonic distortion and 3rd harmonic distortion are both 0.3% or less.
Their percentages are almost equal to each other.
Last edited: