And yes, you'll need to dig that scope out. ;-) You want to do some stability tests to make sure the amp doesn't break into oscillation in the high frequencies. You'll have to experiment with the low-pass filter formed by C1 and R5. You could start with the standard Williamson values of 4.7K ohms and 200pF. With the Heyboer "Peerless" transformers on my amps, I use 1800 ohms and 1200pF. The amp starts to roll of at 30K but it's VERY stable and sounds clean and smooth in the highs. But every transformer is different.
Thank you Grover !! 😀 You are right, i'm gonna do simple with the PSU and with the chokes and big caps lines seem to be already enough filtered
And the first draw of the PCB 🤓
Regards
And the first draw of the PCB 🤓
Regards
Last edited:
That looks nice! May I suggest that, if possible, you add standoffs for the octal sockets? It will keep heat away from the board and reduce strain when changing tubes. My boards were designed for Belton PCB sockets with 15mm standoffs.
Also, in the schematic you have 550VDC caps for the power supply, but on the board it says 500VDC. That's cutting it a bit close with 480VDC. Just checking. ;-)
Also, in the schematic you have 550VDC caps for the power supply, but on the board it says 500VDC. That's cutting it a bit close with 480VDC. Just checking. ;-)
Thank you !! PCB design is my favorite part 😀
So i made a lot of change 😆, first I followed your advice's, i added holes for the standoffs socket around the power tubes and preamp tubes, about the 550V caps don't worry it's because they are not on the board (chassis mount) look closer there is 3 spots for them top right, top left and mid top.
I optimized all the wiring spot's to limit as much as possible the length of them, I added a chassis GND, a spot for the FB (i forgot it previously)
PCB:
It's gonna be 2mm thickness, 2 OZ copper and ENGI
Traces Width:
Power lines and grounds for KT88 are 2mm (max 140mA per tube)
Power lines and ground for preamp are 1mm (max 10mA)
Signal lines are 0.350mm
Filaments KT88 are 2mm (1.6A per tube)
Filaments Preamp are 1mm (600mA per tube)
I ran some math, I can go lower for all the traces because of the 2 OZ copper, what do you think of this:
For the KT88 filaments: 1.7A => 1mm ??
For the 6NL7: 600mA => 0.5mm??
For the KT88 High Voltage : 140mA => 1 mm??
For the signal => 0.250mm??
Space with copper is 1mm, for this parameter i should do 0.254mm of space with all the signals traces to focus them better but it represent so so so much more work 😵
Next step check and check and check, I'll add more ventilation to the board with holes
V2 Rounded Version:
I put 4mm eyelets, 3mm traces for the KT88 filaments, I can maybe go for 4.5 /5mm eyelets but no more, there is no space. They will be twisted like you but separate KT together and 6SN7 together.
I changed again a lot of things and I think I'm not far away from the final version, I played with copper and power lines / filaments or far away from it and the signal is close 0.254mm to focus it, i'm use to do this on guitar amp and it work very well.
V3 Rounded:
What do you think of this ?
By any chance, do you have a website in Europe to buy a Chassis ? I only have access to Hammond but they are not so so nice looking 😆
https://www.tube-town.net/ttstore/en/chassis-al-1444-17133.html
I changed again a lot of things and I think I'm not far away from the final version, I played with copper and power lines / filaments or far away from it and the signal is close 0.254mm to focus it, i'm use to do this on guitar amp and it work very well.
V3 Rounded:
What do you think of this ?
By any chance, do you have a website in Europe to buy a Chassis ? I only have access to Hammond but they are not so so nice looking 😆
https://www.tube-town.net/ttstore/en/chassis-al-1444-17133.html
Last edited:
@6A3sUMMER @trobbins @ra7 @alexcp
Gentlemen,
Could I get your attention on this tube project?
2 of you for the tube circuit design.
2 of you for the PCB layout.
Do you see anything that needs to be changed?
Gentlemen,
Could I get your attention on this tube project?
2 of you for the tube circuit design.
2 of you for the PCB layout.
Do you see anything that needs to be changed?
Orion24, I can offer a few subtleties for you to ponder over.
The 1uF and 100nF coupling caps are contingent on the OPT's primary inductance - they may be larger or smaller depending on performance needs - and you could end up using caps with different lead spacings. C3,C4 should have stray capacitance to gnd minimised - ie. avoid having gnd plane underneath.
The 1 ohm KT88 current sense resistors could be a leaded resistor, such as a 10 ohm 250mW, which could act as a poor mans fuse for protection - so just needs pads added within smt pad.
Each KT88 screen may need a stopper, to avoid hf oscillations and/or poor man fuse protection. That resistor could be on the pcb or directly to the pcb pad.
The feedback lead perhaps should have some trace/pad spacing to gnd, to avoid stray capacitance. In general that may be advisable for most other main signal high-impedance traces which have nearby gnds.
360Vac pads/traces on pcb have greatest working voltage between them, so that creepage distance needs to be more than 360 to gnd.
The CT connection should be direct to C10 neg, and then star out to C11,C20 etc. The neg of C11, C20 should link to the amp pcb gnd for the 1R senses. I'd suggest doing a gnd connection sketch to appreciate if you have any loops and how to handle loops in two amp channel configurations.
For the power pcb, in general its best to star from cap pads, not bring in a resistor connection mid-way between cap pads, so as to avoid trace sections with different currents in them.
Your heater circuit shows a basic humdinger. That's not on a pcb so free to modify, eg. to a pot.
And lastly, I have been using Patrick Turner's current feedback technique of inserting a small inductor in the speaker gnd line, which would be included eg. in your R3 connection to gnd. That variation requires some advanced testing of high-frequency response to confirm ok operation, so may be too difficult to try and include for future proof use.
Given you are planning a Williamson variant, there may be some relevant background information in this link:
https://dalmura.com.au/static/Williamson design info.pdf
Ciao, Tim
The 1uF and 100nF coupling caps are contingent on the OPT's primary inductance - they may be larger or smaller depending on performance needs - and you could end up using caps with different lead spacings. C3,C4 should have stray capacitance to gnd minimised - ie. avoid having gnd plane underneath.
The 1 ohm KT88 current sense resistors could be a leaded resistor, such as a 10 ohm 250mW, which could act as a poor mans fuse for protection - so just needs pads added within smt pad.
Each KT88 screen may need a stopper, to avoid hf oscillations and/or poor man fuse protection. That resistor could be on the pcb or directly to the pcb pad.
The feedback lead perhaps should have some trace/pad spacing to gnd, to avoid stray capacitance. In general that may be advisable for most other main signal high-impedance traces which have nearby gnds.
360Vac pads/traces on pcb have greatest working voltage between them, so that creepage distance needs to be more than 360 to gnd.
The CT connection should be direct to C10 neg, and then star out to C11,C20 etc. The neg of C11, C20 should link to the amp pcb gnd for the 1R senses. I'd suggest doing a gnd connection sketch to appreciate if you have any loops and how to handle loops in two amp channel configurations.
For the power pcb, in general its best to star from cap pads, not bring in a resistor connection mid-way between cap pads, so as to avoid trace sections with different currents in them.
Your heater circuit shows a basic humdinger. That's not on a pcb so free to modify, eg. to a pot.
And lastly, I have been using Patrick Turner's current feedback technique of inserting a small inductor in the speaker gnd line, which would be included eg. in your R3 connection to gnd. That variation requires some advanced testing of high-frequency response to confirm ok operation, so may be too difficult to try and include for future proof use.
Given you are planning a Williamson variant, there may be some relevant background information in this link:
https://dalmura.com.au/static/Williamson design info.pdf
Ciao, Tim
Last edited:
In the Williamson, the input and PI stages draw about the same signal current but antiphase, and are supplied from about the same 305-315Vdc. However your dropper powering scheme shows the input stage is powered from 230Vdc. You may end up needing to adjust your supply rail for the input stage, such that the PI stage is similarly biased to the original. As such it may be easier to connect R34/R42 from the 'B' node and not the 'C' node.
The chokes L1/L2 appear to be chassis mount. You may want to add pads for a parallel R-C across each choke, to form a ripple trap and allow some noticeable reduction in 2nd harmonic mains ripple voltage.
You also take a risk by not using a substantial choke between 'A' and 'B', although that doesn't affect your pcb.
The chokes L1/L2 appear to be chassis mount. You may want to add pads for a parallel R-C across each choke, to form a ripple trap and allow some noticeable reduction in 2nd harmonic mains ripple voltage.
You also take a risk by not using a substantial choke between 'A' and 'B', although that doesn't affect your pcb.
ElArte,
Which amplifier schematic are you going to build?
The one in Post # 19?
Are you going to use a PCB, or are you going to use point to point wiring?
Any needed modifications on a PCB are more difficult than with point to point.
I seems you are going to build a Williamson circuit, right?
The adjustment of the series RC dominant pole on the input tube plate; the global negative feedback cap; RC coupling values; and the output transformer are interdependent, and are important for low frequency stability, and for high frequency stability.
As to how to do all of that, I have to admit that I never designed a 3 stage amplifier with global negative feedback.
Which amplifier schematic are you going to build?
The one in Post # 19?
Are you going to use a PCB, or are you going to use point to point wiring?
Any needed modifications on a PCB are more difficult than with point to point.
I seems you are going to build a Williamson circuit, right?
The adjustment of the series RC dominant pole on the input tube plate; the global negative feedback cap; RC coupling values; and the output transformer are interdependent, and are important for low frequency stability, and for high frequency stability.
As to how to do all of that, I have to admit that I never designed a 3 stage amplifier with global negative feedback.
Last edited:
To be clear, that's the "American" version. The stages are decoupled all in series. The original Williamson splits the first two stages, 22K to the phase splitter and 30K to the input stage. This results in the voltages you are referring to, Tim.
It's never been clear to me whether Sarser and Sprinkle made a mistake that resulted in running the first two stages at lower voltages, or if it was intentional. I've read that their operating points are perhaps preferable, and I know that sonically I prefer the American version, which is what Heathkit adopted for all their early Williamson models.
The Heathkit W3M is the first identification of idle voltages in that Sarser and Sprinkle adaption of powering, showing input stage anode at 50V (compared to 100V), and PI stage at 200V and 60V (compared to 215V and 105V), with no confirmation of supply rail voltage for the PI stage except that it is likely 200+60=260V (compared to 315V). The W4M is the same except indicating first stage is 54V.
The mid-band signal swing of the first-stage anode and PI stage are quite small, due to feedback, and likely don't approach the load-line limits even at low and high frequencies where feedback has fallen and the first stage gain has come up (although loadlines will not then be straight, so other limits may apply themselves).
The mid-band signal swing of the first-stage anode and PI stage are quite small, due to feedback, and likely don't approach the load-line limits even at low and high frequencies where feedback has fallen and the first stage gain has come up (although loadlines will not then be straight, so other limits may apply themselves).
From a practical point of view, modern PCB tube sockets often have a death grip on tube pins, not help because the connectors are soldered in a fixed position. It is worth considering having a dummy tube in place when fitting them, but I think if I was to design a PCB at some point I would use chassis sockets and have holes in the PCBs. I think that simplifies the heater wiring as well, and gives options to mix and match tubes where heater voltages and pinouts are not the same.