The supply rail is split by a diode and a resistor, so the VAS and the front end gets a separately filtered supply. It is the 10R in series with a diode, which feeds the supply rail after the split.
On VSSA board, it is a 0.75W SMD.
On my board, it is R11 and R12, which I specified as 0.5W, assuming a slower rise time at the main supply. (Paulo has just shown us we can get good quality resistors in the same physical size at 1W, and IvanLukic is using MO's for all values). I did not anticipate someone using this kind of supply with my board.
The resistor is the smaller part of the problem. If the rise time on a supply can be as short as LC said, then we have 35V, across a 1/2 watt resistor, feeding a 1000uF capacitor. So for a brief time, the R and the C will see up to 3.5 amps...
Kang Naf pm me too on monday
Pete,@John Bali: you know what they say, "if you never break anything, it just means you are not trying hard enough"
I believe you try much harder than I did
finaly you did that test, thanks...
my Peeceebee use 1/4 watt resistor for all, have good safety information here
cause I also did burn that 10 ohm resistor but still sounding at output(with higher offset & speaker connected)
thats strange isn't it
after replace the 10 ohm everything go back & measured the burned 10 ohm, my meter say 1k ohm
Last edited:
Yes, it is probably metal film, so it burned until the resistance got high enough to limit the current...
"It's only fun until someone lets the smoke out... then, it's hilarious!"
(but only if you have spare resistors, or spare car engines...
)The supply rail is split by a diode and a resistor, so the VAS and the front end gets a separately filtered supply. It is the 10R in series with a diode, which feeds the supply rail after the split.
On VSSA board, it is a 0.75W SMD.
On my board, it is R11 and R12, which I specified as 0.5W, assuming a slower rise time at the main supply. (Paulo has just shown us we can get good quality resistors in the same physical size at 1W, and IvanLukic is using MO's for all values). I did not anticipate someone using this kind of supply with my board.
The resistor is the smaller part of the problem. If the rise time on a supply can be as short as LC said, then we have 35V, across a 1/2 watt resistor, feeding a 1000uF capacitor. So for a brief time, the R and the C will see up to 3.5 amps...
Would it help to lower the value of C5and C6 on your pcb to 470uF? Would it make turn on spike less harming for the resistor? OK, filtration will not be just as good, but probably sufficient...
Last edited:
Yes, but I am reluctant to make that compromise. With a linear supply, the turnon is not an issue. I made some estimes in the beginning, and it seemed like the 1000uF was about the right size for the split rail, for good performance at higher power levels when the ripple on the rails increases.
On power up, the current is always limited by something else - can be the rate the main reservoir caps charge, or an inrush limiter on the transformer, or a series filter resistor in the main supply, etc.
But with a smps, if the design of the smps was to to turn on only after the specified voltage is reached and everything is stable, with 400W smps,
, you WILL have a problem. At the very least, I should add a note to my BOM about the resistor.
Hi all.
Today I started replacing some of the parts with available alternatives and their influence on sonics.
I replaced the 1N4007 diodes with 1N4148. The immediate result was that the muddyness of the mid-high was gone.
Now, this 'muddy' or 'lifeless' symptom appeared with recently installed 35V PS, and a full replace of the capacitor bunch didn't help, the strange anomaly remained and it seemed to become more pronounced in medium to high volume than low volume. With 16V PS of past experiments this didn't occur. My first suspects were those slow diode and the 4 nanosecond glass-dot replenished the sound. Now it's a bit clear to me why LC used ultrafast rectifiers in his boards. Tomorrow I'm gonna buy some MUR160(still 10 times slower than 4148) and test them. Will update when got the results.
Has anybody else here done the same with the diodes? If yes then care to share the findings? Curious...
thanks.
Today I started replacing some of the parts with available alternatives and their influence on sonics.
I replaced the 1N4007 diodes with 1N4148. The immediate result was that the muddyness of the mid-high was gone.
Now, this 'muddy' or 'lifeless' symptom appeared with recently installed 35V PS, and a full replace of the capacitor bunch didn't help, the strange anomaly remained and it seemed to become more pronounced in medium to high volume than low volume. With 16V PS of past experiments this didn't occur. My first suspects were those slow diode and the 4 nanosecond glass-dot replenished the sound. Now it's a bit clear to me why LC used ultrafast rectifiers in his boards. Tomorrow I'm gonna buy some MUR160(still 10 times slower than 4148) and test them. Will update when got the results.
Has anybody else here done the same with the diodes? If yes then care to share the findings? Curious...
thanks.
Correct me if I am wrong, but the diode only comes into it when the cap after the diode is charging, which happens in sync with 2 x line frequency, and does not follow any peak power dips in the rail voltage. The Vdiff should only increase when the rails dip, because 1000u is fairly big in relationship to the power consumption of the VAS, not so?
There is some noise associated with the diode turning off though, and it would be worthwhile to try to improve on that. That noise is in sync with the main rectifier, but does not actually come from there, it is just that the diode follows the ripple voltage. So, a different diode might have better noise characteristics in this application. The noise appears every 1/2 mains cycle, and looks like a small smooth ripple every 8 milliseconds (or 10 if 50 Hz mains f).
I would have thought a schottky diode would have been good there for its speed and lower voltage drop. I've been playing with the idea of bootstrapping the basic PeeCeeBee to maximize voltage efficiency. Sim suggests gaining 10W when using 40V rails and laterals, even more with verticals as in borys thread.
The documentation I received w. the boards from the VSSA group buy specifies 45V rails maximum, for a max power of 100W per channel. In principle, even half of that gives me plenty of dynamic range for normal listening, but it is always fun to see how far you can push before the smoke comes out...
@ Shaan: Interesting that LC also specifies 20V min., so your supply has been just below his spec until now...
LC's pcb uses ALFETs where both output devices are in the same package. That is not good for the reliability but is very practical when you need small pcbs for installing many of them in the same amp case (home cinema, OEM, etc). I think that there will be very much heating at higher powers for a single TO247 package. For that reason LC recommends max 45V. The beauty of PeeCeeBee TH version is that you can have two separate output devices, and even two TO3 devices, so even more power will be available because of better cooling, with PeeCeeBee version than with original LC's boards. With Jason's TO3 version pcb, even 55V should not be a problem. PeeCeeBee may not reach 3MHz but it has it's advantages.
In my boards the caps are 100uF each. The problem occured in upper mids in high volume, with slow diodes, that's all I found.
Probably that turn off noise was causing the porblem. 4148 didn't. Music is very balanced now. Will see what goes with MUR.
Good idea Jason.
@ Shaan: Interesting that LC also specifies 20V min., so your supply has been just below his spec until now...
Yea 4000mV below his specs. And it worked like a charm, every time, with better sonics than a whole lot of low power amps. Not just my conclusion, owners of those amps said it themselves too.
Correct.
Meaning I should change my diodes too?
Not that you "should". But, if faster diodes are available to you then why not? Will wait for your conclusion on the sonic result.
Can't find anything on the former. The BYT is an avalanche diode with 200nS recovery time(50 times slower than 4148).
Sorry, I gave you the wrong part number. It's LQA30T300.
Here's the datasheet: http://www.powerint.com/sites/default/files/product-docs/lqa30t300.pdf
Here's the datasheet: http://www.powerint.com/sites/default/files/product-docs/lqa30t300.pdf
U R correct (tested).
My test pair is repaired and all happy....
(R.I.P. little mosfet, been nice working w. you...
)