I simply soldered one more trim pot on the other side. It was easier than desoldering the one already on the board. I shall make more DC offset measurements the next day.
I also changed 16k resistor to 22k because I intend to use 38,5V rails. Also, when low drop capacitance multiplier is used it will be no less than 38V. I prefer little more distortion than compromised reliability, so front end will probably have a little less current than PMI's amp. I never said that 30mA front end current is error (with 38,5V rails), just that it makes VAS tranies too hot for my taste. In the light of PMI's explanation I think that two pairs of output tranies will require higher resistor value than the single pair version (for the same rail voltage).
I also changed 16k resistor to 22k because I intend to use 38,5V rails. Also, when low drop capacitance multiplier is used it will be no less than 38V. I prefer little more distortion than compromised reliability, so front end will probably have a little less current than PMI's amp. I never said that 30mA front end current is error (with 38,5V rails), just that it makes VAS tranies too hot for my taste. In the light of PMI's explanation I think that two pairs of output tranies will require higher resistor value than the single pair version (for the same rail voltage).
For best performance, the capacitance multiplier power supply I am working on should be adjusted so that the voltage drop across the pass transistor is greater than the maximum expected ripple, at least assuming speakers in the 4-8 ohm range. It does not have to be adjusted that way, but basically I am hoping to reduce the ripple down to a few millivolts.
Where that simply does not meet the needs for one reason or another, it will be possible to use a second identical board, with slightly different component values, connected in parallel off the same transformer/rectifier, and supplying just the input and VAS stage (the part of the amp after the rail diode). Frankly, I hope that will not be necessary in most cases, but it is an option. This way, the voltage for the input and the VAS stage is regulated independently, and with a very small voltage drop. The main supply voltage can be regulated lightly or heavily, as needed.
This all assumes everything works as I imagine...
Last edited:
You're right. It seems the only way to win is to stick all the glasses to one big metal. Ok, this will be done.
(OK, maybe I just have a one-track mind,
I like your mind for that, its safe having your mind around. We need more minds like yours' here. Thank you for your mind-time.
As always, you are too kind...I like your mind for that, its safe having your mind around. We need more minds like yours' here. Thank you for your mind-time.
Pete, I have same thought with you...This all assumes everything works as I imagine...
I want to try increase rails to +/-45V with single pair output but I don't know exactly the value
I must try the highest value resistor first with safety resistor & bulb in series at primary
because small changes in rail then I must check the VAS again...
35V that Shaan suggest is good, with good heatsink too
If you want to do it, you can use this for your starting point.
I drew this peeceebee before you made changes. I strictly followed schematic from post #1, just added one more pair of output transistors and source resistors. Size of peeceebee is 140 x 38 mm (5.5" x 1.5").
Attachments
Thanks Ivanlukic,
I will try 27k at first time & then tweak it a bit...
& I will post here the result
its time to matching some 27k resistors
DC offset:
With both trim pots set to mid position before assy and soldered, offset is very low. It is easy to adjust it to close to zero. I was somewhat misled by PMI’s comment that single trim pot is enough. Indeed, offset is adjusted by turning both trim pots in the same direction, but both must be in circuit. I can not say why both are necessary, but they are. It seems that with symmetrical circuit both halves of circuit ought to be the same. Users of PMI’s pcb design should install both trim pots during assy. This is easiest way to adjust offset, a lot easier than experimenting with offset resistors.
Front end bias:
With 38,5V rails, 22k resistor is too much. Total current for VAS+input is now 9mA, which is low, but I shall leave it this way. Ideal would be 19-20k resistor, but I was too impatient to wait, so I used 22k that was at hand. This will be low front end bias special edition of VSSA PeeCeeBee.
With both trim pots set to mid position before assy and soldered, offset is very low. It is easy to adjust it to close to zero. I was somewhat misled by PMI’s comment that single trim pot is enough. Indeed, offset is adjusted by turning both trim pots in the same direction, but both must be in circuit. I can not say why both are necessary, but they are. It seems that with symmetrical circuit both halves of circuit ought to be the same. Users of PMI’s pcb design should install both trim pots during assy. This is easiest way to adjust offset, a lot easier than experimenting with offset resistors.
Front end bias:
With 38,5V rails, 22k resistor is too much. Total current for VAS+input is now 9mA, which is low, but I shall leave it this way. Ideal would be 19-20k resistor, but I was too impatient to wait, so I used 22k that was at hand. This will be low front end bias special edition of VSSA PeeCeeBee.
I think I am with you on that.Pete, I have same thought with you...
I want to try increase rails to +/-45V with single pair...
On one hand, 35V rail voltage is more than enough, and one of the great things here is that the amp sounds consistently good at lower and higher power (louder listening level or w. lower impedance speakers).
On the other hand, LC has specified his modules to 45V max, and 100W max output... I guess I am like most people, I would like to have 100W, even though I know I don't really need it... and If I can get that from one pair of mosfets, even better...
I am not sure what the limiting factor is, but certainly the input and VAS bias has to be adjusted for that.
If you looked at my chassis above, it is pretty obvious that I am taking higher power and more heat into account.
I made space for two permanently mounted trim pots for what I think is a good reason (and because I was already planning on a larger board). This is for convenience, and ease of adjustment. For best results the bias and offset should really be adjusted twice, once when the circuit is built, and again after a bit of run time with the intended heatsinks, chassis, etc.
But - I made sure my board can be set up exactly as described by Shaan in one of the first posts, with the same procedure, one multimeter, and one trim pot. After setup and some measurement, one single fixed offset trim resistor can be substituted (tested). The is not the easiest way, but it is consistent with the OP.
However, you are correct, I should and will post more specific directions by the time the next revision is done.
edit: did you also measure the total idle current?
No. I was happy to observe that the big heatsink is not hot, only warm, so I skipped that part. I tested with heatsink in upright position as it will be on 3mm thick alu chassis. I noted that when paper is under the heatsink, the temperature is higher, but not much higher. It is obvious that 3mm alu chassis takes some of the heat, serving as additional heatsink. With 16k resistor I was unable to keep my finger on the small VAS transistor for more than a second. With 22k, I do not feel any heat on the small heatsink (no more than 30C on the heatsink). I guess that 9mA for VAS and input transistors will be enough to keep them in linear zone and keep distortion low. I do not have equipment to prove that.
I was surprised to see that offset went down from 260mV (unadjusted) with single trim pot) to 4-5mV (unadjusted) with second trim pot just soldered. All pots were turned to approx mid position before soldering. This means that my little theory about different VAS tranies hfe as culprit went to smoke.
I was surprised to see that offset went down from 260mV (unadjusted) with single trim pot) to 4-5mV (unadjusted) with second trim pot just soldered. All pots were turned to approx mid position before soldering. This means that my little theory about different VAS tranies hfe as culprit went to smoke.
That you add one more trimmer give better balance to the 15-22k "partner" resistor...I changed & swapped 2 x 15k resistor to have little DC offset,
+ side 15k & - side 15k seems to work revert if we decrease the value on + side then DC off going to +....
I've measure 15k resistor about 14,86 ohm changed to 14,94 (rail was +/-33VDC)
that little but not that litle
so far, those resistor really can change the DC offset much... not the VAS not the output pair maybe not the input pair too(nobody try it yet).
I have problem to get BC550C-560C that are closely matched,
I already get them finally
(BC550C-560C unmatched), but the HFE are around 600+ and around 500+...
Will try it & let you all know
maybe we can have something new. Now what I'm thinking is make a new Peeceebee that can safely operate at +/-45VDC or more
cause I don't want to screwed up my first made again , no more sparkling or smokeYes, I have the impression that if I made the PeeCeeBee without any of the trim pots symmetry would be better than with only one trim pot.
My peeceebee has no trimmer on it, I just matched almost all the resistor that more than 470 ohm, the higher one need more attention... (but less than 470 ohm also I've measured)
Shaan do even better no trimmers for adjust & no caps on VAS ,
he tell us some good tips but sometimes we miss it
Ivan,
now you must measure it one by one (that trimmer & resistor)
if you want to replace trimmer with one or two resistor
don't move them off board at same time, you know why right?
But before you do that just enjoy the music for some time,
let it sing & see the behavior maybe need to adjust the trimmer again a little
just in case
Do you have the two boards already done?
Shaan do even better no trimmers for adjust & no caps on VAS
, he tell us some good tips but sometimes we miss it
Ivan,
now you must measure it one by one (that trimmer & resistor)
if you want to replace trimmer with one or two resistor
don't move them off board at same time, you know why right?
But before you do that just enjoy the music for some time,
let it sing & see the behavior maybe need to adjust the trimmer again a little
just in case
Do you have the two boards already done?
Last edited:
One trim pot at mid-travel gives you 500K||15k, or ~14.6 on one side, and 15K on the other. With a gain of 22x, that is enough to account for the offset you measured. This is one reason why I do not use direct coupled input with my own system.I was surprised to see that offset went down from 260mV (unadjusted) with single trim pot) to 4-5mV (unadjusted) with second trim pot just soldered. All pots were turned to approx mid position before soldering. This means that my little theory about different VAS tranies hfe as culprit went to smoke.
When I said one trim pot is enough, I should have also said you may need a fixed R equal to 1/2 of the trim pot value on the other side (sorry).
The unmatched hfe in the VAS transistors should not change the offset, or only a tiny bit. It will have more influence on distortion, through its effect on the positive and negative halves of the signal. We are all in the same boat in that respect. The "E" and "D" parts cannot be matched, and no version of BD139/140 has specs to impress.
I can get some complementary "E" pairs, but there did not seem much interest in better VAS transistors, and there seem to be alternative solutions. Also, this means we are no longer using common and low cost parts, one of the original goals of this thread.
In any case, matching the input pair seems more important, and easier to accomplish.