I'm looking at building a Voltage Controlled Current Source; approx 15V output, with up to 70A available - I need to be able to control the current.
I've found this schematic
formatting link
and if I were to uprate all the components (e.g. from 1/4W resistors to 1W or higher), and replaced the IRFZ48Vs with the IRFZ48Ns which have a higher current load, would it work? I tried to build + simulate this circuit in EWB, but it didn't work out
- but that could be down to my VHDL :)
I've been working with electronics etc for years, but never ventured far from digital solutions, although I have knocked up power supplies in the past, they have never been VCCS.
Correct; planning on building it up as multiple large heatsinks, which are actually copper water cooled.
The PSU will usually only run at around 12 - 15V @ 40 - 50A (so peek of
750W), although I need the extra capacity to be available. I'm also planning on expanding the system as time goes on with additional units to increase the current availability.
The PSU's are to be used for electroplating aluminium at the company I work for, and they will be expected to run for around 8 hours a day under the above load (another reason why I'm over-specing).
You should look around for a "carbon pile". It's just a big huge honking power resistor that you adjust by cranking down on a pressure screw - one of those should have no problem handling only 70 amps. As long as your supply is stable, you should be able to set it and forget it.
Plus, I'd have to have this corroborated by somebody who knows their elbow from a SMPS - but would PWM control work for plating?
Since you didn't say you needed more than one nor how good it really has to be:
How about just buying an off line switcher power supply that has a remotely programmed current limit and running it at the current limit all the time? I'd look for a product like:
Thanks for the advice so far - its appreciated :) The schematic is getting there - just one last question.
I've come across a lot of advice to put a resistor and capacitor in parallel to the tansformers primary coils in order to help regulate the fluctuations between the mulitple transformers - however, how do you calculate the required values?
I'm going to be using 4 x 500VA tordial transformers, with 240V primary windings, with two secondary windings of 25V @ 10A. The secondaries will be connected in parallel to provide the required amperage, and there will be two 56000mF electrolytic capacitors to smooth the output connected in parallel. The primaries will be connected in parallel to the same 240V A/C source (suitably fused etc).
Any advice on the method to calculate the required resistors + caps on the primary windings ?
I thought it was a bit insane too, but hey, I've never parallelled transformers, so ....
Already am :)
Very - see above :) Thought I'd post, just to make sure though, as I was planning on leaving them out. Just didn't want to possibly risk damaging the tordials.
Does plating need filtered DC? You might consider a big transformer driving a packaged SCR bridge rectifier puck thingoe, with a primary circuit breaker and a huge output series resistor as the ultimate protection/current limit. That would be simple, low loss, and practically indestructable.
The multiple-fet thing could turn out to be an immense headache for the relatively inexperienced... envision coffee cups full of exploded fets and plating lines down for days at a time.
It's really hard to design your own gear and come out ahead of a commercial product. If it's expensive, there's usually a reason.
I wouldn't build my own DVM, oscilloscope, TV, or bench power supply... It's just not worth my time, usually by a huge factor.
We see a lot of posts here by people, usually beginners, who think they can build one-piece electronics cheaper than they can buy it. I think that's seldom true.
We plate a lot of different finishes, including chrome but also basic dyed anodisation - the best results always come from a fairly ripple free supply - any ripple changes the thickness of the layer, and although we are talking microns, it can make a massive impact on the end result.
I did get a quote for a unit for this. And it was over a thousand pounds. With a 5 month lead time. The cost could be bearable, but the lead time certainly is not. And we wouldn't have as much control over the system - it could end up that we add additional production lines, tanks etc, so need more PSU's - rather build them in house instead of coughing up large sums and waiting an eternity.
That's a TO-220 rated at 202 amps DC, 333 watts dissipation. If the OP (or anybody else) believes these numbers, he's in for some serious pain.
Note the derating factor of 2.2 watts/degC. Then note the thermal resistances. Then read the footnotes. Why does IR *do* this? Do they want their fets to explode?
TO-247's, directly bolted to a water-cooled sink with silicone grease, no insulator, not too close together, might prudently be used at 100 wats each or so.
Just apply time and money.
It would be a serious project. SCRs would be simpler and a lot safer. There are cheap trigger boards around, packaged SCR bridges, stuff like that. 3-phase would be nice... no filters.
Start at $1686 because the slope front jobs have no filtering at all; they're just xfmrs, variacs and rectifiers.
None of these are voltage controlled AFAIK. They're all adjusted by turning a knob.
Here is a suitable transformer for $110:
formatting link
Add rectifiers for maybe another $100 (being generous)
formatting link
The brute-force water-cooled MOSFET regulator would do the rest. IRF1404 is rated at 162 amps at 40 volts, $3.80 ea from Digi-Key. . Thermal considerations would require using several of these in parallel so figure 30 to 50 bux worth of MOSFET's. We're now up to $250 -- time spent screwing around being free, of course. Control cctry would be trivial, perhaps another $25. The OP already contemplates the humongous heatsink.
Oh, I forgot filter caps. Threephase would be a lot better here because then you could skip the filter caps though you'd need more (but smaller) transformers and more rectifiers.
Figure at least 100,000 uF of capacitor here. 68,000 uF at 25 WVDC are $21.43 ea at Digi-Key. I'd go with several 22,000 uF 40 WVDC caps at $18.77 ea. to handle the ripple current better. 10 of 'em (220,000 uF total) are about $175. Total now is still
I stipulated that heatsink-related issues were "given" per OP's post, and that cost of labor was not included. I think that's the OP's intent: to use available in-house labor to reduce cashflow cost.
Fooey! Please say more about how SCR's would be safer -- or simpler for that matter. I don't think so. Please show sources and prices for the cheap trigger boards and packaged SCR bridges that you mention -- and how these might be assembled to meet the OP's request.
Even with 3 phase excitation there's no way they'll strip ripple like a series-pass bank of MOSFET's would. That flat isn't on with SCR's.
This is not a "serious project" if the OP can do the watercooled heatsink he said he could. Please show how doing it with SCR's would be simpler, even with significantly inferior performance.
Andy, where are you located? I'll be in London soon, would a visit be easy and possible? I'm a retired EE and not looking for work, slug lazy and proud of it -- but I'd like to see your plating op because I'm interested in plating. I do a bit of that, just small stuff in my garage, no chrome. I'll be in London to visit with new granddaughter -- but I can only take so much baby stuff before I start getting antsy. I'd love an excuse to hop on the tube or a train for a day if it's not too heroic an adventure.
You want a design? I'll need a purchase order. If you can build these for $500 each, and the market price is 3x as much, you should go into the business.
I've made a couple of VCCS type plating power supplies capable of reverse pulse waveforms (or other waveforms if you wanted...), but none as big as what you're suggesting. The first one I did used a TDA2030 audio op-amp (very cheap opamp, puts out at least an amp but has large DC offset, and not unity gain stable). The circuit ran off split supplies, probably about
+/-10V. I connected the supply pins of the op-amp to the bases of 2N3055 and 2N2955 transistors to boost the output current, so when the op-amp would start to struggle with the output current, it gets a bit of 'help' from the power transistor. I think it would do about 10 Amps comfortably but I can't remember and didn't need to push it too far anyway because the object being plated was small. (The emitter of the PNP 2N2955 goes to the positive supply, the collector goes to the op-amp output, the positive supply pin of the TDA2030 goes to the base of the 2N2955, and there is a resistor maybe 10 Ohms from the base to emitter. Same for the NPN 2N3055, but the emitter goes to the negative rail and the base goes to the negative supply pin of the TDA2030. The output of my boosted power op-amp went (through a small resistor or inductor to help stability I think...) to one electrode of the plating tank, and the other electrode of the plating tank went through a current sensing resistor to the common of the power supply. I say common rather than ground, because I think I actually floated the power supply itself to allow one of the electrodes to be grounded if so desired. Of course the current sense resistor develops a voltage which is fed back to the inverting input of the TDA2030, and the non-inverting input is driven with the waveform which you want the current to follow, that waveform being with respect to the common of the power supply, (not necessarily ground). I think I used a 555 timer and a 4066 chip and several pots, to make the reverse pulse waveform. I think I needed some fancy things like Zobel networks to make it all stable, most of which I copied out of the TDA2030 datasheet (which contains only audio circuits), and tweaked by trial and error, with various test loads and looking for ringing on a scope.
By the way, the TDA2030A and 2040 are probably better, I think one of them includes protection diodes to the supplies which are absent from the 2030, and the current rating might be better.
The second plating supply was much fancier, it had 16 channels to drive 16 independent plating outputs, and the current sources were proper ones which could drive a real ground referenced load - I used a current sensing resistor in series with the output pin of the power op-amp. I think each channel would drive 1Amp and it was all software controlled using DACs and a PIC microcontroller connected to a PC through a serial cable. If you're really interested I could try to find the old schematics, but it's in protel format so it'd take me a while to fix up a PC to convert it to anything else.
First off, just want to thank everyone for their input; been unwell, so haven't been online a lot recently (hence my dissappearance).
Chris;
Many thanks for sharing your experiences :)
If you would be willing to share, I'd certainly like to take a look! I work a lot with the Atmel AVR Risc uC's at work, and have been drafting including one in the PSU - specifically to digitise the regulation and allow monitoring and control via our network (already have TCP/IP Stacks for the AVR, as well as code that makes use of the built in ADC
- both from existing projects).
Dan; Based a little bit further up the country these days (although until 8
- 9 months ago, I lived just south of London!). We are in the Scottish Borders, about an hour south of Edinburgh. If you want to come up, let me know - am sure I could arrange for a visit! You never know, I might even have the new supply up and running..
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.