I use one of the industry standard Farnell bench power supply.
Works up to 30 volts and a couple of amps.
Mine is about 15 years old and was second hand when I bought it and still going strong.
Works up to 30 volts and a couple of amps.
Mine is about 15 years old and was second hand when I bought it and still going strong.
Mines just a single but if I remember right you can get dual ones too.
www.farnell.co.uk
They have 73 dual power supplies and 939 single power supplies.
On mine earth is on a separate banana socket so you can stack supplies.
www.farnell.co.uk
They have 73 dual power supplies and 939 single power supplies.
On mine earth is on a separate banana socket so you can stack supplies.
Hi Nigel,
Lot's of selection!
When stacking supplies, you have to watch how far away from "ground" your common for each supply can be. IT is easy to exceed that figure if you are not careful.
Another thing to watch is supply stability and output noise (including the broadband noise). The HP supplies excel here. It depends on what you're doing of course. I bought HP supplies due to noise and stability. They were the best. The new ones are really cool. INternal metering is great, and they can log output current/power and voltage. Lot's of other neat tricks they have too! If you're doing basic stuff, not a problem.
Current limiting is another great feature.
Lot's of selection!
When stacking supplies, you have to watch how far away from "ground" your common for each supply can be. IT is easy to exceed that figure if you are not careful.
Another thing to watch is supply stability and output noise (including the broadband noise). The HP supplies excel here. It depends on what you're doing of course. I bought HP supplies due to noise and stability. They were the best. The new ones are really cool. INternal metering is great, and they can log output current/power and voltage. Lot's of other neat tricks they have too! If you're doing basic stuff, not a problem.
Current limiting is another great feature.
If you dont need the zero volts adjustment range, a LM723 plus external NPN darlington is hard to beat for a 2-30V jobbie with current adjustment.
Elektuur (Elektor) designed a very nice lab power supply in 82 with a 723 as floating voltage reference.
Good quality power supplies use an auxillary reference voltage referenced to + or - output, so the error amplifier sees the entire error voltage on its input.
Cheers,
V4LVE.
Elektuur (Elektor) designed a very nice lab power supply in 82 with a 723 as floating voltage reference.
Good quality power supplies use an auxillary reference voltage referenced to + or - output, so the error amplifier sees the entire error voltage on its input.
Cheers,
V4LVE.
There is no shortage of LM723 circuits out there. Some address the 0 VDC limitation, so that technique can be used for whatever supply you choose. You can also make a tracking negative supply to go with the LM723 circuit.
One advantage of the 723 being a very old, well established chip. It blows me away that it is still in production.
One advantage of the 723 being a very old, well established chip. It blows me away that it is still in production.
723 with a clean supply for the internal supply voltage can be really low noise, and much better stability then LM317 and the likes. Check the datasheet, the buried zener is really low noise when bypassed with a capacitor. Theres also a version with bandgap reference but its rarer.
You can even build a pretty sweet shunt regulator with 723.
Heres the manual of the DELTA 030-1. Its a very nice, super quiet linear power supply using Jellybean parts, so you could theoretically build a nice clone with cheap parts. Only odd thing about it is the auxillary supply, they use a 36V 100mA winding to feed a shunt regulator.
Edit, for anyone who wants to build a clone, you can substitute 1n825 with LM329 with a slight change in resistor value and improved noise performance.
You can even build a pretty sweet shunt regulator with 723.
Heres the manual of the DELTA 030-1. Its a very nice, super quiet linear power supply using Jellybean parts, so you could theoretically build a nice clone with cheap parts. Only odd thing about it is the auxillary supply, they use a 36V 100mA winding to feed a shunt regulator.
Edit, for anyone who wants to build a clone, you can substitute 1n825 with LM329 with a slight change in resistor value and improved noise performance.
Last edited:
For tube power amps I find a transformer and a variac work better than a lab supply. At least my experience was that whatever supply I was looking at would either not offer high enough voltage or high enough current or both. For line stages an HP6209 would work well. It's a pretty solid little supply.
For solid state work the main limitation is the output current. I use a pair of HP 6643B power supplies to form up to ±35 V @ ±6 A. But that's only enough for 8 Ω. 4 Ω testing would require 35/4 = ~9 A of output current. With enough added output capacitance it's possible to get that on the peaks, so I'm able to perform the testing I need, but can't push beyond that without getting the 500 W version of those supplies. They're great supplies, though, not cheap.
For higher current I use a pair of Mean Well SE-600-36. That works pretty well but is limited to a single output voltage.
I started out with a supply I bought on ePay or at Bezos' Bookstore. It was around $100 and was a piece of junk. It had a crude resistor ladder DAC in it along with an uC to set the output voltage. It had such poor regulation bandwidth that with a Class AB output stage as the load the output voltage would have several volt of bounce on it. It went back to the seller.
Tom
For solid state work the main limitation is the output current. I use a pair of HP 6643B power supplies to form up to ±35 V @ ±6 A. But that's only enough for 8 Ω. 4 Ω testing would require 35/4 = ~9 A of output current. With enough added output capacitance it's possible to get that on the peaks, so I'm able to perform the testing I need, but can't push beyond that without getting the 500 W version of those supplies. They're great supplies, though, not cheap.
For higher current I use a pair of Mean Well SE-600-36. That works pretty well but is limited to a single output voltage.
I started out with a supply I bought on ePay or at Bezos' Bookstore. It was around $100 and was a piece of junk. It had a crude resistor ladder DAC in it along with an uC to set the output voltage. It had such poor regulation bandwidth that with a Class AB output stage as the load the output voltage would have several volt of bounce on it. It went back to the seller.
Tom
Hi Tom,
For sure (the cheap supply you bought for $100). But really, what did you expect?
I use an HP 6209B. I also have a Xantrex XKW-600-1.7 (600 VDC @1.7 amperes) on the bench. Enough juice for you? lol! Also,Heathkit and a few other HV regulated supplies.
For signal stuff, you can't beat the little HP 6236 or 6237 supplies. The 6255 will allow you to power up most AB transistor amps, and even run them a little. But for full power you really do need a supply just as you would build it in the finished product. So an amplifier power supply off a variac works well. I don't like combining power supplies. Too easy to goof, and having them track can be an issue. I did build some custom bipolar supplies that were variable, regulated. They work, but are ugly and the HP stuff has lower noise.
For sure (the cheap supply you bought for $100). But really, what did you expect?
I use an HP 6209B. I also have a Xantrex XKW-600-1.7 (600 VDC @1.7 amperes) on the bench. Enough juice for you? lol! Also,Heathkit and a few other HV regulated supplies.
For signal stuff, you can't beat the little HP 6236 or 6237 supplies. The 6255 will allow you to power up most AB transistor amps, and even run them a little. But for full power you really do need a supply just as you would build it in the finished product. So an amplifier power supply off a variac works well. I don't like combining power supplies. Too easy to goof, and having them track can be an issue. I did build some custom bipolar supplies that were variable, regulated. They work, but are ugly and the HP stuff has lower noise.
I have two setups. In NJ I use the HP6271's bought surplus when working with SS power amps. I usually run the outputs through the Tektronix DVM's for current. For opamps and lower voltage I use a TEK PS5010, also bought surplus.
In OH I use (as Chris first mentioned) a variac and a big Antek transformer. That power supply is housed in an old HP "Digital Voltage Source" carcase.
For tubes, I use a Heath IP17, or a Variac/transformer combination.
In OH I use (as Chris first mentioned) a variac and a big Antek transformer. That power supply is housed in an old HP "Digital Voltage Source" carcase.
For tubes, I use a Heath IP17, or a Variac/transformer combination.
Attachments
I had expected it to work as a power supply. It was advertised as one...
Those are indeed awesome. The more modern versions (E36xx) are excellent too. I was able to catch and repair a couple out of the surplus pile while I was at National Semi/TI.
Tom
Nice setup Jack!
Hi Tom,
True, you expected them to work properly. But these days anything goes. They even use the DAC in an MPU to decode streaming audio. It doesn't work well, but it works and they sell it anyway. Same for really cheap power supplies and other instruments.
I have an E36312A and EDU36311A. The E36312A is painfully slow for button responses. The EDU36311A is much better, but still slow. My EDU36311A locks up on a Chrome page as well. They are really nice supplies though. I love that they are ethernet connected, too many things are USB and my USB port hub is already full at 7 positions. The wonderful thing about the two supplies is that you can set current limits at low values, like 10 mA for example. Sometimes I do have to limit that low.
Hi Tom,
True, you expected them to work properly. But these days anything goes. They even use the DAC in an MPU to decode streaming audio. It doesn't work well, but it works and they sell it anyway. Same for really cheap power supplies and other instruments.
I have an E36312A and EDU36311A. The E36312A is painfully slow for button responses. The EDU36311A is much better, but still slow. My EDU36311A locks up on a Chrome page as well. They are really nice supplies though. I love that they are ethernet connected, too many things are USB and my USB port hub is already full at 7 positions. The wonderful thing about the two supplies is that you can set current limits at low values, like 10 mA for example. Sometimes I do have to limit that low.
Yeah...
What works for me really well is the CLC front end (small values C, no paralleling), followed by a series regulator (2 X TO3)... for about 7A of constant current. Dropping 10V across... hence a nice chunk of a heatsink. But OMG... it is just simply wonderful. L effective above 100kHz, liner regulation below 100kHz... the sound of silence.
@v4lve lover You may want to check this thread...
https://www.diyaudio.com/community/threads/post-you-smps-noise-spectrum-measurements.394518/
What works for me really well is the CLC front end (small values C, no paralleling), followed by a series regulator (2 X TO3)... for about 7A of constant current. Dropping 10V across... hence a nice chunk of a heatsink. But OMG... it is just simply wonderful. L effective above 100kHz, liner regulation below 100kHz... the sound of silence.
@v4lve lover You may want to check this thread...
https://www.diyaudio.com/community/threads/post-you-smps-noise-spectrum-measurements.394518/
Last edited:
Yeah. I was drooling over that a while ago. Drooling is free, right? I don't need to feed my test equipment fetish. 🙂
Tom
Stacking is great and I found it very very useful.
You need to think how you want use the supply, do you want to be able to perform measurements with low noise?
It’s possible with a SMPS supply but (a) they need alot of post filtering and (b) they’re not as robust/simple to fix as a the linear and (c) their level of limiting often starts in the 100mA range. Trade off is deskspace.
I have two 60V smps cheapo supplies and still managed to get to ~-160dB post filtering with bipolar regulation for an ADC (DC to 100KHz).
But by the time you have that filtering in place, it probably as much deskspace as the linear 🤣
I'm too lazy to build a bench-top SMPS supply -- what Linear Tech (now ADI) demonstrated more than a dozen years ago was that you can control the slew rate of the switch and avoid the rapid di/dt which is the source of the noise spikes -- the tradeoff is somewhat lowered efficiency.
There are mosfet drivers you could put between a PWM controller and the switching fet that allow you to do just that.
If you are too lazy, perhaps you want to give some input into my project? I have found a flyback core HA4061-AL that does 500V nominal 120W 90-250VAC @ 1Khz. Bummer is that it doesnt have an auxilary winding for the primary side controller.
I know how to layout boards, construct a working PSU, its just that im too poor at spice to make sure it is guaranteed to be stable over its entire range.
A box just arrived at my door, 400Pc 39mH 300mA common mode chokes... Now i need a use for them.
Cheers,
V4LVE
If you are too lazy, perhaps you want to give some input into my project? I have found a flyback core HA4061-AL that does 500V nominal 120W 90-250VAC @ 1Khz. Bummer is that it doesnt have an auxilary winding for the primary side controller.
I know how to layout boards, construct a working PSU, its just that im too poor at spice to make sure it is guaranteed to be stable over its entire range.
A box just arrived at my door, 400Pc 39mH 300mA common mode chokes... Now i need a use for them.
Cheers,
V4LVE
Hi V4LVE,
You don't need to use Mosfets for that, BJTs work as well and are easier to (properly) drive.
Hi Jack,
That was a great app note. Efficiency is still much higher than linear, but to get low noise it is far more complicated to construct. In the end, on the bench, low noise is more important as long as regulation is equal between the two types (and it is).
Hi Tom,
It was time to replace the 6236 and 6237 supplies. The new supplies offer some useful features and I needed to rebuild the original HP supplies I used since the 1970s anyway. I still love them, and I will still use them. It's nice to just twist the damn knob to make it work. The biggest advantage with the new supplies is the current limit, and they can be mounted at the far corners of a busy bench and be controlled / read on my computer. The EDU36311A has higher voltages than the 6236/6237 supplies, and for me I needed just a bit higher voltages.
So, if you want logging of current / power or anything, the E363xxA makes sense. it is slow to respond to control inputs (they may fix this issue). The EDU36311A is faster, a little higher output voltages and may be more suitable. But they are low noise. extremely stable and the displays read actual output voltage and current, the set values in lower in the display. Besides, Keysight no longer makes those nice, basic power supplies. At least the new ones do have control knobs that allows you to adjust the output depending on external factors. Setting an exact voltage is of course much easier.
I still have a 6255A and other supplies in that model series. Performance was next to impossible to beat, which is why I still own and use them. But after 50 years in service, they need maintenance and I still need a supply to use while I have them apart for that. These are products where I should send HP a cheque! They have lng outlasted life expectations!
You don't need to use Mosfets for that, BJTs work as well and are easier to (properly) drive.
Hi Jack,
That was a great app note. Efficiency is still much higher than linear, but to get low noise it is far more complicated to construct. In the end, on the bench, low noise is more important as long as regulation is equal between the two types (and it is).
Hi Tom,
It was time to replace the 6236 and 6237 supplies. The new supplies offer some useful features and I needed to rebuild the original HP supplies I used since the 1970s anyway. I still love them, and I will still use them. It's nice to just twist the damn knob to make it work. The biggest advantage with the new supplies is the current limit, and they can be mounted at the far corners of a busy bench and be controlled / read on my computer. The EDU36311A has higher voltages than the 6236/6237 supplies, and for me I needed just a bit higher voltages.
So, if you want logging of current / power or anything, the E363xxA makes sense. it is slow to respond to control inputs (they may fix this issue). The EDU36311A is faster, a little higher output voltages and may be more suitable. But they are low noise. extremely stable and the displays read actual output voltage and current, the set values in lower in the display. Besides, Keysight no longer makes those nice, basic power supplies. At least the new ones do have control knobs that allows you to adjust the output depending on external factors. Setting an exact voltage is of course much easier.
I still have a 6255A and other supplies in that model series. Performance was next to impossible to beat, which is why I still own and use them. But after 50 years in service, they need maintenance and I still need a supply to use while I have them apart for that. These are products where I should send HP a cheque! They have lng outlasted life expectations!
For what it's worth, I'm not trying to use a supply to measure anything. Just to test gear before making a final PS, look at waveforms at a gross level, and just make sure there's nothing terrible going on with a project. But I'm fascinated by the discussion and will keep reading.