Here's a design I came up with out of necessity. It's simple/cheap/quiet/reliable.
I had designed an A/D acquisition sensor for my company that could measure voltages up to 60Vdc. I needed to test it with a known V source . It was difficult to find a finely adjustable bench DC source that would go that high without spending a lot of money so I threw this together and have since realized how useful it is for other projects. For example this can be used to supply power for any preamp design using a single supply. One could replace the pot with a fixed R.
It's probably not original but I did come up with it on my own just from past design experience.
The front end is a conventional toroid transformer/rectifier/filter. The output circuit implements a capacitance multiplier topology, which is essentially a transistor configured as an emitter follower with an RC to the base. The Hfe of the transistor effectively multiplies the capacitance. In the end it's as if you put the equivalent cap on the output of the rectifier. In this case I used a power darlington TIP142 with a gain over 1000. 1000 x 10uf = 10,000uf. So I have this appended to the front end rectifier filter which already has an 8200uf cap. Hence, It's ultra quiet with no visible ripple at half load which is about 1/4 amp.
The darlington is current protected using one of my favorite constant current circuits. Essentially a small signal transistor will turn on, shorting out the base emitter junction of the darlington, turning it off, when the 1 ohm series sense R develops 0.65v, ie 0.65 amps. It probably wouldn't survive indefinite duration short circuit due to heat/dissipation. One could always attach a simple thermal switch to the transistor heatsink. It would open up the secondary DC. The heatsink in this case is a piece of 1/8" thick 2 x2 angle aluminum that also doubles as a front panel.
Using a 10 turn panel mounted pot the output V can easily be adjusted in 10mv steps.
It is unregulated, If you need precise regulation add a voltage reference device and an opamp as the source for base drive.
I built a prototype with components on perf board mounted on a piece of 1/4" plywood. (Oh no! Call NFPA). I used proto board and a spring contact terminal block.
The PCB uses about $15.00 dollars worth of parts. External parts total about $45.00, the most expensive items being the transformer/10 turn pot/binding posts.
25 circuit boards from Exprss PCB would be about $325.00. If you eliminate solder mask and silkscreen about half that.
If there's any interest, I will put up a zip file with schematic/BOM/Express PCB layout file/Tina simulation.
Here's the schematic and photo anyway.
![IMG_0454[1].JPG](https://www.diyaudio.com/community/attachments/img_0454-1-jpg.1008427/ "IMG_0454[1].JPG")
I had designed an A/D acquisition sensor for my company that could measure voltages up to 60Vdc. I needed to test it with a known V source . It was difficult to find a finely adjustable bench DC source that would go that high without spending a lot of money so I threw this together and have since realized how useful it is for other projects. For example this can be used to supply power for any preamp design using a single supply. One could replace the pot with a fixed R.
It's probably not original but I did come up with it on my own just from past design experience.
The front end is a conventional toroid transformer/rectifier/filter. The output circuit implements a capacitance multiplier topology, which is essentially a transistor configured as an emitter follower with an RC to the base. The Hfe of the transistor effectively multiplies the capacitance. In the end it's as if you put the equivalent cap on the output of the rectifier. In this case I used a power darlington TIP142 with a gain over 1000. 1000 x 10uf = 10,000uf. So I have this appended to the front end rectifier filter which already has an 8200uf cap. Hence, It's ultra quiet with no visible ripple at half load which is about 1/4 amp.
The darlington is current protected using one of my favorite constant current circuits. Essentially a small signal transistor will turn on, shorting out the base emitter junction of the darlington, turning it off, when the 1 ohm series sense R develops 0.65v, ie 0.65 amps. It probably wouldn't survive indefinite duration short circuit due to heat/dissipation. One could always attach a simple thermal switch to the transistor heatsink. It would open up the secondary DC. The heatsink in this case is a piece of 1/8" thick 2 x2 angle aluminum that also doubles as a front panel.
Using a 10 turn panel mounted pot the output V can easily be adjusted in 10mv steps.
It is unregulated, If you need precise regulation add a voltage reference device and an opamp as the source for base drive.
I built a prototype with components on perf board mounted on a piece of 1/4" plywood. (Oh no! Call NFPA). I used proto board and a spring contact terminal block.
The PCB uses about $15.00 dollars worth of parts. External parts total about $45.00, the most expensive items being the transformer/10 turn pot/binding posts.
25 circuit boards from Exprss PCB would be about $325.00. If you eliminate solder mask and silkscreen about half that.
If there's any interest, I will put up a zip file with schematic/BOM/Express PCB layout file/Tina simulation.
Looks pretty good to me! 🙂 Just a couple bits . .
- safety-wise, it's likely wrong to share the Ground with Line-neutral -- sooner or later you'll encounter a mis-wired outlet with disastrous consequences; other folks on here know better than me, though, so maybe wait 'til those smarter chime in
- 1,44 / 0,36W (latter if only 60V at the rectifiers) might a little high for a ten-turn pot . . 25 or 50k would still provide enough current for the Darlington
(- couldn't say at C1, 'cause you have two C3's instead 😉 )
- if it was me, I'd include a 1k or so resistor in series with Q1's Base . . on a bench supply, the 2N5210 is bound to see a capacitor-charging spike some time
Cheers
- safety-wise, it's likely wrong to share the Ground with Line-neutral -- sooner or later you'll encounter a mis-wired outlet with disastrous consequences; other folks on here know better than me, though, so maybe wait 'til those smarter chime in
- 1,44 / 0,36W (latter if only 60V at the rectifiers) might a little high for a ten-turn pot . . 25 or 50k would still provide enough current for the Darlington
(- couldn't say at C1, 'cause you have two C3's instead 😉 )
- if it was me, I'd include a 1k or so resistor in series with Q1's Base . . on a bench supply, the 2N5210 is bound to see a capacitor-charging spike some time
Cheers
Last edited:
That ground is necessary to make simulation work in Tina since I was going for a comparison of AC in vs DC out. I should have posted the non- simulated version of the schematic. It doesn't have it, rather the third prong GND is connected to the secondary return.
Yeh ...two C3s?
I have also since changed out the 2N5210 for a MPSA06 when I looked at the max V ratings while making a BOM. (After I posted). Note in the picture there's a surface mounted transistor, FZT493, which is a 100 volt device. I wanted to not use SM for we hobbyist's sake.
I've never considered putting in a base resistor in that config. The constant current attribute has always worked for me without it, but its' use in the past was usually with regulated supplies - in a circuit quite downstream, so probably a good precaution to do so. I'll have to experiment to come up with a value. Currently that current protection is damn fast and consistently saves the day.
Currently I've never seen the rectified/filtered V get above 46 volts. ~ 0.2watts on the pot. Perhaps for the sake of longevity it would be a good idea to raise that value. I just happened to have a beautiful 10k pot with integral ball bearings in my junk drawer.
Thanks for the input.
See attached update.
Yeh ...two C3s?
I have also since changed out the 2N5210 for a MPSA06 when I looked at the max V ratings while making a BOM. (After I posted). Note in the picture there's a surface mounted transistor, FZT493, which is a 100 volt device. I wanted to not use SM for we hobbyist's sake.
I've never considered putting in a base resistor in that config. The constant current attribute has always worked for me without it, but its' use in the past was usually with regulated supplies - in a circuit quite downstream, so probably a good precaution to do so. I'll have to experiment to come up with a value. Currently that current protection is damn fast and consistently saves the day.
Currently I've never seen the rectified/filtered V get above 46 volts. ~ 0.2watts on the pot. Perhaps for the sake of longevity it would be a good idea to raise that value. I just happened to have a beautiful 10k pot with integral ball bearings in my junk drawer.
Thanks for the input.
See attached update.
Attachments
Last edited:
Oh yeah -- if your junk drawer has a ten-turn with ball bearings, hay-al yeah -- design that puppy in!
Now I understand 😉 !
Cheers
edit: Sorry, I didn't notice the FZT493. Still say the proto is beautiful! By the way, the higher gain, lower noise of the 2N5210 surely has benefits. And the 2x Vbe of the TIP142 will hold the 2N5120's Vce to a few volts, even if you do fit a large-ish Base resistor 😉
Now I understand 😉 !
Cheers
edit: Sorry, I didn't notice the FZT493. Still say the proto is beautiful! By the way, the higher gain, lower noise of the 2N5210 surely has benefits. And the 2x Vbe of the TIP142 will hold the 2N5120's Vce to a few volts, even if you do fit a large-ish Base resistor 😉
Last edited:
Years ago, a friend built a similar power supply based on, I think, LM723. But the first time he shorted the output, the transient surge destroyed the current limit transistor so we added a ~100 Ohm resistor in series with the current limit sense input. After which it survived being shorted, no problem. But today when I search for the LM723 schematic, it already has such a resistor on chip??? So, I have to wonder if National, etc had the same problem and integrated the solution? If you look up most any implementation of this current limit circuit, you will see a base resistor for this reason.
Ok, I'll put a 100 ohm R in there. I can easily simulate that and check if it has any bearing on the effectiveness of I limit.
A 100W incandescent bulb has around 140 ohms, might be nice to see what’s going on, unless you have a meter on there somewhere.
Heh... Reminds me of the power supplies I built when I was in 5th/6th grade. I learned a thing or two about limiting the base current in BJTs in those designs. Turns out R1 is mission critical... 🙂
Neat li'l project.
Tom
Neat li'l project.
Tom
+1
Adding 1 kΩ should have only negligible impact on the current limit. Q1 shouldn't draw much base current.
You may also want a reverse biased 1N4007 or 1N5408 across the output just in case the load does something nasty like providing an inductive kickback.
Adding a large resistor from the pot wiper to ground could be a nice fail-safe in case the wiper ever lets go. 100 kΩ should be good.
Oh... And a diode from output to input to discharge any load cap when the power is turned off. Without such diode you could end up with quite a bit of reverse voltage across C-E on the pass transistor.
Tom
I knew the cold resistance of incandescent lamps is dramatically lower than the operating resistance, so I found a 100W lamp and measured it. The result: 10 Ohms. HP used this effect for automatic gain control in their original sine wave oscillator.
https://www.hpe.com/us/en/about/history/innovation-gallery/002-product.html
In 1973 I tried a 120V 60W lamp and found 16Ω, cold. Which is exactly proportional to your data (the 4% is negligible slop).
I loaded the 16Ω terminals of a 15W tube amp with the 120V 60W bulb and got impossible numbers. And a very dim glow. By 10 or 20V the filament was already over 50Ω.
I’d put a switch on the filament bulb(s)so it can be bypassed, but would be nice to have various options for the output r and have it all in one apparatus. That would make for a few less burned parts in the things I build.
It's not terribly difficult to make the current limit variable, but then maybe you want to handle transient current spikes and maybe not. How fast do you want the limit to be enforced? Sinclair amp power supplies in the 1970's had a folding protection that shut down when they were overloaded. This was simply done by using the output to bias the reference, and a RC network for startup. Today switch-mode supplies are common but maybe not at higher voltages. An advantage of PWM is that you can convert a limited current at the raw voltage into a much higher current at a lower voltage, vs a linear supply wastes a lot of heat when the raw voltage has to be dropped to a low voltage.
https://www.eleccircuit.com/variable-switch-mode-power-supply-circuit/
https://www.eleccircuit.com/variable-switch-mode-power-supply-circuit/