New 183 build

[Stephen1966]

positive feedback is good! negative is a problem

We can take it further, by asking ourselves, if it is only about positive feedback and it is good, why doesn't he apply to both stages? Why only the outgoing stage?

on the picture it is on both stages, and also i see on many D pictures repeatable detail - from speaker wires to PI second plate, so on all 3 stages?

Repeated yeah. It's part of his method. But instead of saying "outgoing stage," it would have been better to say the second stage of each tube (the b section).

[erwin_ve]

Actually the plate of v1a is in phase with the cathode of v1b. Same with v2a and v2b. Positive feedback.

Agreed. But the cathode of V1a with the cathode of V1b are they also in phase? They aren't physically tied together, (that is, they are designed as ordinary - anode follower - inverting gain stages) and induction is a relatively small force, but even small changes in the cathode affect big changes at the plate. I'm also thinking what happens in a LTP where the [PI] cathodes are tied together in the tail - the biasing effect it produces.

[Edit: I see what you are saying about the phase relationship of the plate wire in this arrangement - agreed. And apologies it was a poor use of words in my post. I was just thinking about the effect of the cathodes together... but you're right, there is also the effect it has on the signal in the plate as well.]

No worries! Off course cathodes together is negative feedback.
Looking at several pics of ODS , there is a always positive feedback going on. On virtually every stage, even the output stage.

[WhopperPlate]

I have heard this story echoed here before:

Started out with lead dress in a amplifier with power tube grid wires twisted together and ran to grids through a typical Marshall path , under the PS board and between the power tube sockets .

Eventually I untwisted them and moved them to the traditional Dumble arrangement: paralleled with OT secondaries next to v3.

The difference in characteristics was startling. Went from a harder to play sound to a softer looser vibe. Case in point .

Such differences are considered less than other aspects. One might spin in circles attempting to dial out a frequency response that is a direct result of lead dress .

Eliminating these interactions is ultimately not the objective. on the contrary , it’s all about creative utilization through understanding .

[Stephen1966]

Back to the subject of pots...

In keeping with the original 183 there are just 4 pots which present a builder's problem.

B350k BASS
A270k TREBLE
A272k (OD) DRIVE
A285k (OD) VOLUME

Scraping the traces of the pots is something I am going to try and so for each of these, I will be starting with a 250k pot. It looks problematic from the outset though. First, I am a human with fat fingers and a lazy brain, so achieving any kind of consistency when removing material from the traces is going to be hard. How hard, time will tell, but I am more likely to achieve this with the BASS pot because of its linear taper. The log pots though will present a different order of difficulty because if I understand the manufacturing process well, the taper is achieved by overprinting. So the carbon deposits aren't going to be an even depth on the phenolic wafer. I might reach the desired resistance but completely destroy the taper in the process.

So, as a backup I am also looking at using parallel resistors on the pots instead. Starting with 500k pots and adding parallel resistors to drop them down to the desired resistances. I've used Tony Bone's spreadsheet, from the link I posted earlier and I've just posted the schematic for my build in the first post of the thread with the parallel resistors I calculated.

https://ampgarage.com/forum/viewtopic.p ... 72#p444772

Almost any parallel resistor value is going to induce a taper (log or anti-log) off the original taper. The spreadsheet uses the model of a linear taper but it can easily be used to judge the effect on log tapers, if not exactly then enough to see where the values are drifting and an impression of how much.

Starting with the B350k BASS

This is a B500k pot with a 1M2 CC resistor in parallel

350k.jpg

(The text up the side should read 'Percent Division')

Things get trickier now because there are limited values in the available CC resistor stock, but going with what's available we can get pretty close to the target resistances.

A270k TREBLE - 264K with 500k//560k

264k.jpg

A272k (OD) DRIVE - 276k with 500k//620k

276k.jpg

A285k (OD) VOLUME - 288k with 500k//680k

288k.jpg

As you can see, the taper becomes more linear, the higher the parallel resistors value but even with the 264k value (with the deepest taper) the taper is around 40% and less than 10° off centre.

That's all in an ideal and perfect world where components come as advertised. And where the tolerances in manufacture are in the double figures I may be lucky and find a combination of pots and resistors that get me somewhat closer to the mark. If I'm really lucky I will be supplied pots that are really off the mark and which don't require any parallel resistor shenanigans or scraping - Go figure Those close to 250k are more likely to be near the mark but the BASS pot is almost certainly going to need to be a product of modification.

Anyway, I posted my schematic in the first post and there are some other changes in there you might spot as well. If you see anything that looks wrong, please let me know. I welcome the feedback.

Stephen

[WhopperPlate]

I almost never find a dead on spec pot. buy 10 and you will probably get lucky .

Honestly , I know you are hesitant to scrape the trace on the potentiometers, but you are going to be surprised how little it takes to go from a 250k to 300k . This is why I always hook up alligator clips to take constant measurements during the procedure . You can easily overshoot the mark. Only scrape just hard enough .

Opening then up is arguably harder . A small flat head to wedge under the tabs is all you need to get them loose.

[Stephen1966]

I have heard this story echoed here before:

Started out with lead dress in a amplifier with power tube grid wires twisted together and ran to grids through a typical Marshall path , under the PS board and between the power tube sockets .

Eventually I untwisted them and moved them to the traditional Dumble arrangement: paralleled with OT secondaries next to v3.

The difference in characteristics was startling. Went from a harder to play sound to a softer looser vibe. Case in point .

Such differences are considered less than other aspects. One might spin in circles attempting to dial out a frequency response that is a direct result of lead dress .

Eliminating these interactions is ultimately not the objective. on the contrary , it’s all about creative utilization through understanding .

This doesn't surprise me at all and I feel sorry for all those point to point guys out there staring at their bowl of spaghetti and wondering where it all went wrong No sympathy at all actually. But lead dress is only one contributing factor. The same principles can be seen in his component placement on the boards. Take a real good look at his degooped preamp sections. Bear in mind here, the goop would have had a mechanical purpose: holding the components firmly in position. And then, around the chassis, consider his ground points and the sections of the amp that connect up with them, where they are. Consider also the outside foil placement on the caps. 6PSs are relatively noisy compared to DMEs which barely show a flicker when you test them side by side. Is anyone ever going to spill the beans on cap placement? Consider he used solid core in his heater wires, mitigating any oscillation in them.

We need to study everything minutely if we are going to get anywhere near that next level where he operated. That and considering his meticulous attention to detail we have to overcome the temptation of thinking any of this was the result of an accident (happy or otherwise) and be open to the idea that there is often more than one reason why he did anything.

[bepone]

A270k TREBLE
A272k (OD) DRIVE
A285k (OD) VOLUME

Scraping the traces of the pots is something I am going to try and so for each of these, I will be starting with a 250k pot.

maybe no need to do that.. you can install regular 250k pot, and if missing 20-30k , add in value with the existing resistor going to the pot.. if the resistor is 100k, you put 130k, if the resistor is 150k, put 180k....

[WhopperPlate]

This doesn't surprise me at all and I feel sorry for all those point to point guys out there staring at their bowl of spaghetti and wondering where it all went wrong

Or what went right!!!! lol I have played more than a few rats nests that were somehow magical . Marshall’s come to mind .

…but then again Marshall’s aren’t known for their pristine hi fi clean tones….

I know it’s as redundant and tired as saying “tone is in the fingers” (it is … but these are guitar amplifiers, and especially when it comes to overdrive (not limited), theoretical ideals fly out the window.

[WhopperPlate]

A270k TREBLE
A272k (OD) DRIVE
A285k (OD) VOLUME

Scraping the traces of the pots is something I am going to try and so for each of these, I will be starting with a 250k pot.

maybe no need to do that.. you can install regular 250k pot, and if missing 20-30k , add in value with the existing resistor going to the pot.. if the resistor is 100k, you put 130k, if the resistor is 150k, put 180k....

If I understand what you are saying correctly, adding a resistor to either end of a pot is going to create a fixed value you can’t change, limiting the range of the sweep. Doubtful that’s desired .

[bepone]

think two times , what is the practical loss of 250k-270k? pot rotation loss now to 9.5 instead of 10/10? Drive and OD vol are usually on 40-50% and resistor is there already..

[Stephen1966]

think two times , what is the practical loss of 250k-270k? pot rotation loss now to 9.5 instead of 10/10? Drive and OD vol are usually on 40-50% and resistor is there already..

And think again

The problem isn't really the sweep of the pot but, I think two-fold...

First, to take a 250k pot and bump it up to 350K you are attenuating the signal on a side by side comparison with a 250k.

The problem with using only one parallel resistor - to give you a desired taper, is that while it augments the signal in a 500k to a 350k translation, the other side of the wiper still sees the original impedance. So in a linear pot at 50% rotation on the side with the parallel resistor you have (350/2) = 175k, on the other side of the wiper you have an impedance/resistance of (500/2) = 250k. It's the combination of both these factors, the signal strength on one side and the impedance/resistance on the other which give the pot its special "flavour".

To get around it, We might place two parallel resistors either side of the wiper; so for 350k (actually 352.692k) using a 500k pot, we might use two parallel resistors of 1M2 each. It gives this taper:

350k (2).jpg

[bepone]

The problem isn't really the sweep of the pot but, I think two-fold...

First, to take a 250k pot and bump it up to 350K you are attenuating the signal on a side by side comparison with a 250k.

a no ,small misunderstanding, i didn't want to touch 350k, only 250->270 or 280k, on drive and OD volume pots..there is serial resistor before and just needed to increase value on resistor.. for 350kohms probably i would do 500k pot and parallel from the outside..

[Stephen1966]

The problem isn't really the sweep of the pot but, I think two-fold...

First, to take a 250k pot and bump it up to 350K you are attenuating the signal on a side by side comparison with a 250k.

a no ,small misunderstanding, i didn't want to touch 350k, only 250->270 or 280k, on drive and OD volume pots..there is serial resistor before and just needed to increase value on resistor.. for 350kohms probably i would do 500k pot and parallel from the outside..

No, actually, I think we are on the same page... I used the 350k example because it is the most exaggerated of the problem pots. Same principles apply to the smaller 270/272/280k pots though. True, you could add fixed resistors in these nets. You could in the example of the OD, change the 100k after the coupling cap to 120k, and change the second from 180k to 220k but putting them on the hot side of the pot like that, we would be attenuating the signal. So instead we could use smaller additional fixed resistors on the ground side of the pots and then, we could use more exact values to get closer to our desired pot values; adding a 22k to our drive pot and a 32k to our Volume pot. This would affect the sweep of the pots to the extent that you could never turn them fully down and that might be acceptable. It would lift/add impedance as a consequence as well. Likely as not, it wouldn't make much difference to the way we use them because we might never use such low settings for Drive and Volume, or the highest settings either; the usable ranges are likely to be around the middle.

It may be a moot point with these near-to-250k pots anyway. Personally, pots which are over the stated resistance values are rare in my experience, with pots that are under, being more common. But right there, straight out of the bag, they might be close enough to use without any additional changes. I'm not holding my breath though!

Yours is a possible solution (and I thank you for it) but my first choice would still be to take a 250k and scrape the traces. This seems the only way preserve all the elements without cost. I would get the right level of signal attenuation and the right level of impedance to ground in the signal chain. My next attempt, if that one failed would be to use 500k pots with two equal value parallel resistors (effectively wired in series) one on the CW lug to the wiper and one on the CCW lug to the wiper, as this seems the only way to preserve both the signal strength and the impedance to ground. So there are three possible approaches here if we don't count the idea that they will be ready to use out of the bag, because let's face it, that's never going to happen is it? Anything that gets me within, say 5% of the target value or better yet, within 2% is probably going to be just fine. I might encounter other problems in the circuit when it comes to tuning it, but with tolerances like those I will be able to rule out any pot foul play.

In any case, it seems that if scraping the 250k up to 350k doesn't work I will have no choice but to use the parallel resistor method there. To tag a 100k in series with a 250k pot, anywhere in the chain or net, would have marked consequences.

[bepone]

ok for the pot modification

i just tried.. it is going very fast, 3 min job with dremel style cutting disc.
from 250k i made 350k.. slightly taken external envelope 0.5mm around the carbon track..

IMG_20221003_145944__01.jpg

IMG_20221003_145948__01.jpg

[Stephen1966]

ok for the pot modification

i just tried.. it is going very fast, 3 min job with dremel style cutting disc.
from 250k i made 350k.. slightly taken external envelope 0.5mm around the carbon track..
IMG_20221003_145944__01.jpgIMG_20221003_145948__01.jpg

Okay! Super! And that's an Alpha pot is it. I know the traces are easier to get at with a CTS.