Old school power supply with Pi filter
Ash Small, Wed Feb 13 2019, 10:41PMSo, just finished the power supply for my spring reverb.
I didn't want to use silicon for voltage regulation, I wanted to do it 'old school'.
I wound a choke using 0.5mm magnet wire and a couple of 60mm type 26 double thickness cores, which came out at 125mH (1/8 Henry), and made a Pi filter using four 4700uF electrolytics, 1 to 3 ratio, as appears to be recommended, and checked via simulation (Duncan amps simulator and LTSpice).
Transformer is rated 27V @ 54VA, and tested at 0.3A (300mA) is giving 36V.
Voltage is stable to within 1mV...
I know most will think I'm crazy doing it this way, but I didn't want the hassle of silicon voltage regulators.
The spring reverb circuit has been tested using another supply at 38V, and draws just under 200mA from that, so voltage with this supply will probably rise a bit, I may have to 'waste' some power somewhere to bring the voltage back down to ~35V..... I'll test the whole thing tomorrow, it's getting late here now....
Re: Old school power supply with Pi filter
2Spoons, Wed Feb 13 2019, 10:54PM
You may have 1mV short term stability (nicely done btw!), but you are now totally reliant on the grid for long term stability. Which is usually +10% to -15% from nominal. I have no idea if that matters in a spring reverb.
2Spoons, Wed Feb 13 2019, 10:54PM
You may have 1mV short term stability (nicely done btw!), but you are now totally reliant on the grid for long term stability. Which is usually +10% to -15% from nominal. I have no idea if that matters in a spring reverb.
Re: Old school power supply with Pi filter
johnf, Thu Feb 14 2019, 09:01AM
2spoons i can echo that!!!!
johnf, Thu Feb 14 2019, 09:01AM
2spoons i can echo that!!!!
Re: Old school power supply with Pi filter
Proud Mary, Thu Feb 14 2019, 12:11PM
I always use linear supplies for HT+ with a C-L-C pi filter.
I was surprised at the small inductance used in your supply.
As a ballpark figure, tried and trusted over generations, values for a 250V/75mA supply would typically be 8μF - 8H - 16μF.
Proud Mary, Thu Feb 14 2019, 12:11PM
I always use linear supplies for HT+ with a C-L-C pi filter.
I was surprised at the small inductance used in your supply.
As a ballpark figure, tried and trusted over generations, values for a 250V/75mA supply would typically be 8μF - 8H - 16μF.
Re: Old school power supply with Pi filter
Conundrum, Fri Feb 15 2019, 04:53AM
Incidentally how feasible is it to mod a 110V SMPS for 220V? The component that failed was the capacitor which incidentally was Crapxon.
Somehow the original owner managed to find a lead that fit the back and KABOOM! He said it was plugged in for less than 300ms as it blew the plug fuse.
So far the transistor and coils look OK but input filter is underrated.
Most of the parts are fine and I did try the PSU on a 110V input and 47uF cap and it worked for a short time probably because the console wasn't hooked up to video.
-A
#include "yay_my_lcr_returned.h"
Conundrum, Fri Feb 15 2019, 04:53AM
Incidentally how feasible is it to mod a 110V SMPS for 220V? The component that failed was the capacitor which incidentally was Crapxon.
Somehow the original owner managed to find a lead that fit the back and KABOOM! He said it was plugged in for less than 300ms as it blew the plug fuse.
So far the transistor and coils look OK but input filter is underrated.
Most of the parts are fine and I did try the PSU on a 110V input and 47uF cap and it worked for a short time probably because the console wasn't hooked up to video.
-A
#include "yay_my_lcr_returned.h"
Re: Old school power supply with Pi filter
Ash Small, Tue Feb 26 2019, 12:43AM
Tho photo didn't load, I gather there's a problem somewhere, hope it gets sorted soon.
In response to your comments, that is how the maths works out, but we all know the formula is a function of inductance and capacitance, if you reduce one be a factor of 100 you have to increase the other by a similar factor. I'm using 4,700uF for the resevoir cap and three times that for the smoothing cap.
The references I found seemed to suggest a 1:3 ratio was best, rather than 1:2 as in the 8uF:16uF example you gave, which is typical for 'old school' power supplies.
The caps I'm using are heavy duty industrial grade electrolytics kindly supplied by a fellow 4HV'er, stud mount and heavy, I've not actually measured the ESR..... Any suggestions how to do this would be useful
As for the other point raised regarding regulation and stability, I've tested the circuit at 38V for extended periods and at 36V. I've now added a resistive divider for a 5-6V supply (not sure yet if I'll use it for an IC), total resistance is 75 Ohms, and this drops the voltage to between 31V and 32V (the biggest swing I've seen is 0.6V), this allows ten percent safety margin for the 35V rated caps.
The JFET's I'm using are only rated for 25V, but they don't tend to see full rail to rail voltage, and I chose them carefully, and I've tested the circuit at 38V for extended periods.
I tend to think it's voltage spikes that kill transistors, or overcurrent. I've managed to kill one so far, at the time I was seeing 700mA on the DMM.
I know back in the day the top of the range amps like Standel had voltage regulation gas tubes rated at 150V, but most didn't, and mains voltage was more variable back then. The most variation I've seen here while testing is 2%.
EDIT: I'm drawing 600mA on the divider if I recall correctly, and just under 200mA on the circuit itself, which consists of eight JFET's in class A toplology, one of which is using ultrafeed topology, and pumps anything up to 100mA into the reverb tank itself.
The choke barely gets warm.
Ash Small, Tue Feb 26 2019, 12:43AM
Proud Mary wrote ...
I always use linear supplies for HT+ with a C-L-C pi filter.
I was surprised at the small inductance used in your supply.
As a ballpark figure, tried and trusted over generations, values for a 250V/75mA supply would typically be 8μF - 8H - 16μF.
I always use linear supplies for HT+ with a C-L-C pi filter.
I was surprised at the small inductance used in your supply.
As a ballpark figure, tried and trusted over generations, values for a 250V/75mA supply would typically be 8μF - 8H - 16μF.
Tho photo didn't load, I gather there's a problem somewhere, hope it gets sorted soon.
In response to your comments, that is how the maths works out, but we all know the formula is a function of inductance and capacitance, if you reduce one be a factor of 100 you have to increase the other by a similar factor. I'm using 4,700uF for the resevoir cap and three times that for the smoothing cap.
The references I found seemed to suggest a 1:3 ratio was best, rather than 1:2 as in the 8uF:16uF example you gave, which is typical for 'old school' power supplies.
The caps I'm using are heavy duty industrial grade electrolytics kindly supplied by a fellow 4HV'er, stud mount and heavy, I've not actually measured the ESR..... Any suggestions how to do this would be useful
As for the other point raised regarding regulation and stability, I've tested the circuit at 38V for extended periods and at 36V. I've now added a resistive divider for a 5-6V supply (not sure yet if I'll use it for an IC), total resistance is 75 Ohms, and this drops the voltage to between 31V and 32V (the biggest swing I've seen is 0.6V), this allows ten percent safety margin for the 35V rated caps.
The JFET's I'm using are only rated for 25V, but they don't tend to see full rail to rail voltage, and I chose them carefully, and I've tested the circuit at 38V for extended periods.
I tend to think it's voltage spikes that kill transistors, or overcurrent. I've managed to kill one so far, at the time I was seeing 700mA on the DMM.
I know back in the day the top of the range amps like Standel had voltage regulation gas tubes rated at 150V, but most didn't, and mains voltage was more variable back then. The most variation I've seen here while testing is 2%.
EDIT: I'm drawing 600mA on the divider if I recall correctly, and just under 200mA on the circuit itself, which consists of eight JFET's in class A toplology, one of which is using ultrafeed topology, and pumps anything up to 100mA into the reverb tank itself.
The choke barely gets warm.
Print this page