Agreed, my plan is to post an alternative layout to etch as soon as I test the circuit, which will take 2-3 days. I
I already have one, but no point in posting something I have not tested.
The diodes are in a straight line, to use a single flat heasink if desired for low power. For much higher power dissipation in the rectifier, the diodes and pass transistors mount from the bottom, also in a straight line, at 90-degree bend, and attach to bottom of chassis, or a flat piece of aluminum, or any cheap heatsink you have on hand. So, always only cheap heatsinks needed, or a bit of extra work. No extra expense for more continuous power output when used with high bias (or higher overall power like VSSA circuit w. double-die Exicons).
Hopefully Marc can help w. single layer, b/c I do not usually etch boards.
Using 8A/1000V KBU8M bridge rectifier (suitable for pcbs) would simplify things considerably and give small footprint on the pcb with enough space for some kind of heatsink. But I like PMI's board because fast discrete diodes can be used, if someone prefers using them.
Attachments
Last edited:
Yes, and it can make it easier to route the connections in a small area. One reason why I used the old TO-220AC package for the diodes is because they were used in the original circuit posted above. Another reason is it gives us some flexibility to experiment with different types of diodes.
The main filter and reservoir cap is intended to be 10,000uFd, 35-mm diameter, snap-in leads. The holes will also accept 12.5-mm spacing straight lead (checked already), at the expense of the snap in leads being a bit loose.
The output cap would depend on the application, and on what kind of rail caps are on the amplifier board. The board will accept a 4700uFd 25-mm diameter cap, or any standard smaller, 16-mm diameter elco.
In the case of the VSSA modules, which have 1000uF per rail, most of the current for a good transient response will come from the amplifier itself, so a small-ish output cap on the power supply may be enough. The low on-resistance of the pass transistors makes a lower value/lower cost output cap possible.
I will post some pictures over the weekend to illustrate what I mean, showing the main components. I tried hard to use only parts that are easily available.
The outside dimensions are 120x95 mm
The mounting holes are on 110x85 mm centers, 3.5 mm diameter, unplated. 6/32 or 3 mm mounting screws. 5 mm minimum standoff (spacer) between circuit board and chassis, assuming untrimmed snap-in lead on the larger caps.
At the end of the board where the rectifier diode heatsinks would mount, there is a second set of mounting holes 7 mm in from the first set, in case of heatsink and mounting screws overlap.
Pete,
There is one unexpected benefit from your modified version of Mr. Evil's capacitance multiplier circuit - j-fet to ground instead of opposite rail. If one needs a single rail supply for amplifiers like Rod Elliott's DOZ, only positive side of the pcb should be populated with components and you have single rail capacitance multiplier.
Interesting, I had not thought of that, but you are correct, the two circuits are separate now (with the exception of the bridge)....If one needs a single rail supply for amplifiers like Rod Elliott's DOZ, only positive side of the pcb should be populated with components and you have single rail capacitance multiplier.
In any case, I know you have been waiting, so here is the progress so far:
Circuit board inspected, and one minor error found in mounting holes for the pass transistor heatsink (not electrical).
One board assembled, dropout voltage set , and powered up with lamps as a load. Seems to be working so far, ripple is vanishingly small, nothing seems to overheat and everything looks stable so far.
Attachments
For me, no
...I like to put things into inaccessible spaces... sort of striving for the apple-inside-the-brandy-bottle look, and the LEDs inspired me.... Also, I had some other constraints on length of the board, b/c I am working on a "compact" chassis. The board is designed allow the diodes to mount from the bottom at 90-degree angle, so if desired the six (!) cheap and little heatsinks can be replaced with one large and cheap heatsink (or an even cheaper plate of aluminum).
When mounted on top, the diodes are assembled to the heatsinks first, then the heatsinks with the diodes are soldered to the board, with the board upside down and resing on the heatsinks. (I did not actually have the intended heatsink on hand today, so these are a substitute.) The 10.000uFd caps are soldered last, because they are the largest/tallest part.
So attaching the hatsinks is not a problem, de-attaching them after everything is soldered is another matter, or course.
Attachments
It is not necessary, as long as one remembers that the heatsinks become part of the cirquit and can be shorted with a test lead etc.
The pics are of my first board. I did not want to worry about one more short hazard while testing, so for me, it is a good idea...
I would have normally used screw terminals for testing, instead of Fastons, but I ran out, and did not want to have some of each on one board. Not ALL of the blade connectors (Fastons) need to be populated, in fact I doubt anyone would use all of them at the same time. That was something else useful only for flexibility in testing.
I tested with a small load approximately equal to the VSSA quiescent current, and compared to two other power supplies, w. 37V (35V nominal) rails. Jfet current is 5.7 mA, dropout voltage set to about 0.75-0.80V. I also ran the through-hole version of VSSA with the supply, and there was no hum or buzz, and no noise visible on the output with my scope, even w. very long power supply leads and an open chassis.
This is very approximate, using only my o-scope to measure:
#1, Bridge Rectifier and 2x4700uF/rail, 200mV ripple+ some diode sw. noise.
#2, Bridge Rectifier, 3x4700uF/rail, C-L-C-R-C, ~80mV ripple
#3, Improved Capacitance Multiplier, 10,000uF + 4700uF output cap, <10mV ripple
Basically what I am doing here, is comparing the output of a bridge rectifier with cap filter alone, to two methods of secondary filtering.
#1 is no additional filter. #2 is 0.33R+4700uF, #3 is the power supply I am testing. When I get a bit more more into it, I will post more accurate results and some pics.
Next step is to do a moderate load test at about 1A from both rails. I plan to do this using light bulbs as a load. This is not great for accurate measurement, because they take some time to stabilize as they get hot, but I do not have other suitable loads or a load bank.
Pete, very good work!
My friends and I have built some different amps in the last years and we have always found some improvements making use of cap multiplier PSU instead of standard CRC or CLC PSU, especially for sonic nuances and in the width of music stage.
Till now we've manufactured them in Dalington configuration and with breadboards.
The circuits looks really promising and your boards professionally made.
Please go on with your tests and take me into account in case you decide for a small batch of boards or for a GroupBuy.
Thanks and regards
Nicola
My friends and I have built some different amps in the last years and we have always found some improvements making use of cap multiplier PSU instead of standard CRC or CLC PSU, especially for sonic nuances and in the width of music stage.
Till now we've manufactured them in Dalington configuration and with breadboards.
The circuits looks really promising and your boards professionally made.
Please go on with your tests and take me into account in case you decide for a small batch of boards or for a GroupBuy.
Thanks and regards
Nicola
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.