Circuit Specialists triple power supply teardown

[slburris]

I recently received a triple power supply ordered from Circuit Specialists.
It's 0-30V at 0-3amp X 2 and a 5v @3amp supply:

http://www.circuitspecialists.com/prod.itml/icOid/10338

While I'm guessing it's one of the Mastech variants, I was looking for
a supply with LED displays, not LCD, because I find the LEDs to be
much more readable, especially at funny angles to the faceplate.

One of the things I liked about this supply, is that the current and
voltage are color coded.  While I like shiny things as much as the
next fellow, I can tell at a glance if I'm looking at voltage or current.
The voltages are green and the current is red.  Simple!

Let's look at the front panel.  It's the usual assortment of 4 displays, color
coded as I mentioned, each with current and voltage controls.
In the center is a mode switch which allows each supply to run
independently, or in parallel for 0-30v up to 6 amps, or in series
to run as 0-60v up to 3 amps.  Neat!



OK, let's see how well the voltage displays measure compared to my
Fluke 87-V.  First we have the left and right supplies:




and now the non-adjustable 5V supply.  Hmm, maybe I can find some
adjustments inside to tweak things, not that they are very far off.



Switching to amps, let's short out the supplies and see how the readings match.
Here are the left and the right:




So far so good, next I'll have to pop the cover off and see what we've got
inside.  I'd do that in another post.

Scott

[allanw]

Good stuff. Once you calibrate these power supplies once I've found they typically stay in good calibration for quite a while. There should definitely be some pots in there to adjust. I don't understand why they don't calibrate them better before shipping out though. Maybe too much time spent sitting in a warehouse...

[slburris]

I've now opened up the power supply and I was completely wrong about the
manufacturer.  This supply is made by Atten, and it an OEM version of

http://www.attenelectronics.com/products/dc_power/aps3003d_5d.htm

Let's peek inside.  First the overview.  Big transformer, several PCB with pass
transistors mounted on heatsinks with a cooling fan.  The front panel consists
of 2 PCBs, a display board and an output and pot board.

Large: http://electroscott.com/images/inp1.JPG


The power PCBs look pretty straightforward, relays to select taps on the transformer
and a bunch of analog circuitry.  No part looks unobtainable, good if we need to
repair this.

Large: http://electroscott.com/images/inp2.JPG

Large: http://electroscott.com/images/inp3.JPG


But what's this?  Look in the upper left corner of the last image.  Something
look amiss?  Well it's this:

Large: http://electroscott.com/images/inp4.JPG


Yes, a bridge rectifier has been twisted almost 90 degrees.  This should have
been fixed by QA.  In fact, I couldn't let this go by and I fixed it myself.

Finally, the front panel boards.  Again, nothing too exotic.  The displays
are driven by the ubiquitous ICL7137 variants.

Large: http://electroscott.com/images/inp5.JPG


Scott

[slburris]

Time for more testing.  First we have a series of overshoot tests.
The first diagram is the fixed 5 volt output.  Doesn't look bad at all.
The next two diagrams are at 3.3v and 12v.  Also looks quite good.

The final two measure noise.  The first is the 5v fixed output, showing
about 20mv of noise.  The second is an adjustable set to 3.3v.  This
is much quieter, about 7mv of noise.

All in all, this seems to be a nice power supply. 

Scott

[Mechatrommer]

sorry i am interrupting here, but qurious... why some PSU use those big big transformer, and other use none, just some descreet IC or big opamp/transistor/amplifier only? what the difference in term of efficiency and whats the pro and con?

[saturation]

Scott,

Awesome job! Kudos on your review, it should serve to model what others can do for gear reviews.  I also enjoy seeing all that gear on the floor, as in your pics  Its a good unit, I'd buy this if I needed one.  I current am in the market for one more PSU, so I'll have 3.

The PCB layout looks like its an analog supply, I'm surprised you can see that noise, it won't be significant for the most part, but it could be picking it up from the ICL7107 chips. Its better to test for noise under different load conditions, 3,2,1,0.5 A etc.,

Is the cooling fan always on, or on demand?  Its in a perfect spot regardless.

Fixing that twisted bridge rectifier was a good move, but opening cheap Chinese devices are always a good move to double check their Quality Control.

One performance test I'd recommend is testing the built in voltmeter accuracy under load.  In my Mastech supply, its off by as high as 1-5%, i.e., 3V 1A on the Mastech is really 2.9V 1A, as confirmed by a Fluke voltmeter on the load and another Fluke set up to read amps, simultaneously.  The built in voltmeter reads its worse under maximum load, 3A, 30Vdc.

I like that red/green combo LED, and yes, I think LED look much better overall because they're easy to see an any angle.

[saturation]

Hi shafri,

As a quick eyeball rule of thumb, if you see a large transformer, versus a small transformer with a lot of inductors ferrite rings and beads all over the design, chances are you are looking at an analog power supply, versus a switching power supply.

Analog have cleaner output voltage, but large and heavy, particularly as amperage rises.  They are less efficient, meaning they lose power to heat instead of converting volt/amps as you need.  They are cheaper at < 30Vdc and 3A or so.  As requirements rise, they will cost more than switchers as the transformer starts to get big and expensive.

Switchers can be slim, compact, and more efficient, but are much noisier than analog for any specific output voltage.  They also cost more initially but become cheaper in high power needs.

sorry i am interrupting here, but qurious... why some PSU use those big big transformer, and other use none, just some descreet IC or big opamp/transistor/amplifier only? what the difference in term of efficiency and whats the pro and con?

[slburris]

I should point out that this is a multi-tap linear supply, where each tap
supplies a different voltage so the linear regulators don't have to
dissipate too much power.  For this supply, there is the click of a relay
approximately every 7 volts, 3 clicks total, so there must be 4 voltage taps
on the transformer output. 

Scott

[Zero999]

Yes, that's how it should be done. The trouble is building a similar PSU yourself would involve winding your own transformer, although centre tapped transformers are widely available so having one tap change is easy. Awhile ago I designed a circuit which used a transistor and a comparator to change taps for someone on another forum.

[saturation]

Yep, the Mastech does a similar thing, and it makes things more efficient, as it can be.
Thus, the regulators don't get so hot, when sourcing very low voltages.  I brought it down to 100mV, the lowest it could keep stable, to see how it would handle and it did well.

I should point out that this is a multi-tap linear supply, where each tap
supplies a different voltage so the linear regulators don't have to
dissipate too much power.  For this supply, there is the click of a relay
approximately every 7 volts, 3 clicks total, so there must be 4 voltage taps
on the transformer output. 

Scott

[slburris]

Another way to go about this is to have an adjustable switching regulator
as a pre-regulator followed by a linear regulator.  The switching regulator
is set to output just above the dropouit of the linear regulator.
Then you don't need a huge transformer.  The Hameg HM7044 does this.

The downside is that it's a lot more parts, plus linear regulators don't
do such a good job rejecting high frequency noise, so you have to
pre-filter that as well.

Scott

[Zero999]

Yes, that's called a hybrid regulator and is a very good solution to the problem. It also makes it easier to go down to 0V which is very hard with an SMPS.

[NiHaoMike]

And there's also the trick of using SCRs in a controlled rectifier for preregulation. The main problem is that the high peak currents tend to cause EMI issues. But before cheap high power MOSFETs were available, it was about the only way to build a large power supply. I used one of those ancient monsters in ECEN 441 class to power motors. It was about the size of a microwave oven, weighed 100lbs or so, and buzzed really loudly at high output current. Regulation was poor (had to adjust it after changing the load) but it's not too bad for a piece of equipment as old as Britney Spears. Especially after considering the age of the capacitors inside. I remember it was rated for 100V at up to 10A, or 1kW.

Nowadays, an equivalent power supply would be less than half the size of a shoebox. Times sure have changed!

[cybergibbons]

One of the things that bugs me about my GW-Instek power supply is that only the LED displays are red/green - they haven't followed through the colour coding on the front panel silkscreen or the knobs.

One thing that really annoyed me on a power supply at work was that the outputs (4 of them) were all compressed on the right hand side of the front panel, whereas the controls were spread across. I like to have the outputs directly below the controls.

When you put it into series/parallel, are the displays intuitive? Again, my GW-Instek is a bit strange - in series it will show 28.2/28.1 on the two current displays for a total of 56.3V (always the two displays are 0.1V out), in parallel it will show 2.54/2.59 for a total of 5.13A. It just makes it that little bit harder to read. Supplies I have used before either change the display or show exactly half on each.

[cybergibbons]

Also, that's a lot of noise for a linear supply! The little peaks are very curious. Are you sure it isn't being picked up from something else?

[cybergibbons]

Is the power switch on the front a mains switch, or just an output on off?

Again, from previous experience, the ones where the outputs are switch by the mains input, there tends to be problems with overshoot and nasty spikes when turning on and off. Ones with a secondary output on/off tend to produce cleaner output.

Can you set the current limit approximately using the display with the output off? Or do you need to do it experimentally with a load?

[saturation]

Thanks cg for your insight and comparison, which Instek PSU model do you have? I think you posted it at the forum sometime ago, but having the model #  we can get a fuller comparison.

Is the power switch on the front a mains switch, or just an output on off?

Again, from previous experience, the ones where the outputs are switch by the mains input, there tends to be problems with overshoot and nasty spikes when turning on and off. Ones with a secondary output on/off tend to produce cleaner output.

Can you set the current limit approximately using the display with the output off? Or do you need to do it experimentally with a load?

[slburris]

Is the power switch on the front a mains switch, or just an output on off?

Again, from previous experience, the ones where the outputs are switch by the mains input, there tends to be problems with overshoot and nasty spikes when turning on and off. Ones with a secondary output on/off tend to produce cleaner output.

Can you set the current limit approximately using the display with the output off? Or do you need to do it experimentally with a load?

RE your noise question, I'm going to have to use a better scope than the Rigol to check
the noise out again -- I'm not sure how much to trust the Rigol for low noise measurements.

When in series voltage, the voltage displays show the same number, but you have to remember
to double the value.  Similarly in parallel mode you have to double the current.  If they had used
a micro with some ADCs, displaying the right stuff would have been trivial, but they use the same
3 1/2 digit all in one chips that everyone else uses.

The power switch appears to be wired between the AC inlet in the back and the transformer,
so it appears to be a mains switch, not an output switch.  The wires are routed along the
side of the case and testing did not seem to show any noticeable overshoot on power on.

You have to set the current limit by shorting the outputs -- it would be nice to be able to
just set it, but I haven't seen that feature on low end supplies.  Programmable supplies,
yes, but those cost a lot more.  Are there any supplies of this class which do better?

Scott

[cybergibbons]

Sorry - it's a GW-Instek GPS-4303 - I guess it normally costs about $600 though, but I picked it up for £179 (about $280). I don't think the extra you pay is really worth it.

I think even under full load you'd struggle to see the typical load - I certainly can't on this one. There's slight overshoot if you are drawing the full 3.2A at 63V but it is so small as to be negligible.

This one allows you to approximately set the current limit when off. It always slightly over-reads - so if I set it to 1A, it gives me 0.96A. It's quite useful.

The V displays are within +/- 0.1%, current within +/- 0.2%.

I've also got a PSP-2010, which is a programmable switch mode supply. Noise is far worse, and it's not great at dealing with highly dynamic loads. But it's so much more efficient - the GPS4303 generates a lot of heat when it's just above the tap voltages of 7.5, 15 or 24V.

[slburris]

Looks like the GPS-4303 is going for $399 now in the US.  Very nice
looking quad power supply.

I wonder if my supply is an Atten clone of a GW-Instek model?  They have
copied other OEMs in the past (Rigol).

Scott

[allanw]

You have to set the current limit by shorting the outputs -- it would be nice to be able to
just set it, but I haven't seen that feature on low end supplies.  Programmable supplies,
yes, but those cost a lot more.  Are there any supplies of this class which do better?

I love my TTI PL330QMD I picked up off ebay. Got the dual supply version for $80 and it lets you set the current limit when the output is turned off. And the sense terminals are useful for accurate voltage measurements under load.

http://www.used-line.com/c-415F00555863-aaa.htm