Calibrating the analog voltmeter on a PS

[toli]

Hello guys,

Yesterday I've purchased a used bench power supply. Its not the best in the world, but it'll do for my needs (2 channel with 0-40V and 1A). Its also build quite well (I've had a little look inside). However, one of the voltmeters isn't accurate, and shows about 4-5% higher than the actual voltage (measured with the Agilent U1253A). I would like to recalibrate it so it'll be closer to the actual voltage, but I'm unfamiliar with these analog voltmeter, and can't find the schematic or manual of that power supply online. Can it be done? Can someone point in the right direction?

BTW, other than that it works great. Both channels were tested all the way from 0V (actually about 0.05V minimum) to 40V, and work great even under load. I've also tested the current limiting circuit and it works just fine on both channels. So all that's left to do is calibrate the voltmeter if possible, and its as good as new



Edit:
Problem solved
After reading a bit about these analog voltmeters I've understood that the resistor that's put in series with them it setting the scale. The resistor in these are internal so instead of taking it apart, I've hooked a pot in series, and turned it until the reading consists with my DMM. Thankfully it drifted to a lower value so I can just add a resistor (with a value of 2K) in series to the meter, and there's no need to open the plastic case of the voltmeter to replace the internal resistor.

Edit 2:
In case anyone in the future will read it and will need to recalibrate one of these on their own. The solution I've used (changing the value of the series resistor) is only good if the error is relative to the reading. If there's a constant error, it can be fixed even easier, there's a small plastic screw on the case of the meter that turns the needle. Just turn it a little bit to the direction you need and that's it.

[saturation]

Glad it went well.  It has a nice unique look, not many PSU use those types of dials, its worth upkeep for the look. Did you take pictures when you took it apart and can you share

[toli]

Yes I did. It was full of stickers and markings on the outside, and had quite a bit of dust on the inside, so I took it apart and cleaned it inside out.

[saturation]

Wow.  There's a lot of room inside, easy to repair, when needed.  From the size of the components and room, for an analog supply, it looks like it was designed with a lot of parts derating.  The heat sinks are large by today's standards even for just convection cooling and are they exposed to the outside?  Really easy to stay cool if your lab is comfortably cooled.

I'd be curious to know what its specs are, if you've the time to determine its ripple under maximum load, and output stability is in mV, that is can you dial out 10.00 stably? 10.000, 10.0000 etc.,

[HLA-27b]

Lovely little supply. Never seen any like that.

I think this style of meters never became popular because of its somewhat highish parallax error. On classic meters the plane of the scale and also the plane the needle travels on are parallel and normal to the viewer. On these models the planes are arc shaped and the distance between the needle and the scale varies, being more severe at both ends of the scale, thus aggravating the parallax error. Notice how the upper left meter is not at zero. Anyways, they were superseded shortly by the numeric displays.

In the old days there were some light reflecting meters that projected the shadow of a needle to a semi transparent screen. I wonder how much resolution they got. Speaking of resolution, my professor of history of architecture had a Kern DKM3 theodolite, that baby could resolve 1/10 arc second! 12960000 divisions full circle. I believe they are still worth a few thousands dead.

Oh never mind me just lingering thoughts, I am so distracted....

[toli]

HAL-42b, I've calibrated all 4 dials (2 for voltage and 2 for current) the best I can. One showed relative error, and another one or two had a fixed error (didn't return to zero when it was turned off - a small screw fixes that). However, it indeed has higher error on the edges, but its within acceptable region, especially as its analog and you won't be able to read it as accurate as numeric displays on digital meters.

saturation, the heat sinks aren't exposed,  they are sitting inside the case, which has vent for air flow in all 3 directions around them (bottom, side, top). I did run it on full load (1.1A) into a 15R power resistor, and the heat sinks weren't nearly as hot as I though they'll be.
I'm still a student, and we have quite a few of these PSU in the basic laboratory (into + analog intro labs), and I've always figured they were a piece of s**t. But now that I've purchased one, and had a look inside I'm surprised by the build of it.

As for your other question, I'd be more than happy to do so once I have some time.
Dialing a specific voltage is quite easy, it has a single multi-turn pot. If you are gentle enough you can tune it with about 10mV increments.
Regarding ripple, what would be a good way to test it? What voltage should I turn it to? What load should I put (a resistor, or a CCS like the one Dave had in one of the videos - I have one similar I've build running on a 9V battery and using a PMOS for the output)? I'm not sure my meter is accurate enough to measure small AC riding on DC sources. I've tried measuring my sigma11 (you can look it up on amb.org) which has a very low output ripple and noise, and it reads about the same on both of these (with no load connected). On the very low scale of ACmV range the 1253A isn't accurate enough.

I've just connected an 8R 50W resistor to the supply, and limited the current to 1A (measured with the DMM as well), and connected the DMM probes directly to the output of the amp (to remove and losses or errors due to cables), and measured ACmV across the output. It still reads exactly the same as under no load at all (0.68xmV). Its probably even lower, but as I've mentioned earlier, I get the same reading even when testing the sigma11 which has a much lower noise and ripple (according to measurements made by AMB).

I have the PCB for the LNMP which I'll build sometime in the future, and than I'll be able to make much better readings. The problem is I need to order quite a bit of parts for the LNMP, most of which are available at Digikey, but one of the (the charge controller MC33340P) isn't available there (so if anyone has one he could spare I'll be glad to know), and having one item delivered to Israel is just not worth the cost

BTW, I've just found a couple of these on eBay so if anyone is interested in seeing some more pictures of such a PSU, here's one of them:
http://cgi.ebay.com/HORIZON-ELECTRIC-Model-DHR-40-1-Dual-Power-Supply-230V-/220759657844?pt=LH_DefaultDomain_0&hash=item33664d0d74

[saturation]

Hi toli

Don't worry about whether the dials and the actual voltage are spot on, its not crucial, your DMM is more accurate.   Your dials get within the ball park, but better accuracy can't hurt to save you some time. 

Warm up your PSU to operating temp.

To measure ripple you need a scope and a DMM; place a high wattage resistor across the inputs and dial up a voltage so you get 1A into it.  Confirm the amps and voltage on the load with the DMM.  Place the scope across the leads, if there is ripple you can see it.  The ripple % is ~ ACVrms/DC.

For accuracy, dial down to as close to zero volts you can, and up the maxmium voltage.  Note the reading on your DMM.  The 1253a is good to 5 digits, so you may be able to see if it reads stably down to the uV from less than 1V to mV up to 40V.  If the last digits of your DMM fluctuate, just note where is the most stable reading at , 100uV? 1mV? etc.,



I see what you mean about the heat sinks from the eBay photos.  Enjoy your PSU!

HAL-42b, I've calibrated all 4 dials (2 for voltage and 2 for current) the best I can.

[toli]

Your wish is my command 

Stability:  Once a load is connected (I gave it a few minutes to warm up, driving 1A into 8R) I've measured:
8.4V(1A): Reading is at 8.392 with no fluctuations at all.
0.84V(0.1A): Reading is at 840.1mV with no fluctuations at all.
BTW, the meter is set to Normal in the smooth setting.

Ripple: With the scope set to AC coupling (so I'll be able to measure the AC's Vrms) it reads:
0.1A(0.84V): 800uVrms (Which according to my calculator comet to 0.1% ripple).
1A(8.4VDC): A bit higher with ~1mVrms (which once again according to my calc comes to 0.012% ripple).

I'm not familiar with the numbers I should expect from such a basic and somewhat old PSU, but seems like the numbers aren't bad at, more than worth what I've paid for the unit

[saturation]

Thanks so much toli!  Its a decent, if not good PSU.  For comparison, you can take the specs of an Agilent analog PSU that retails for $450 new.

http://www.home.agilent.com/agilent/product.jspx?pn=e3610a

Or a Chinese PSU:

http://www.mastechpowersupply.com/dc-power-supply/linear-power-supply/mastech-regulated-variable-dc-power-supply-hy1503d-15v-3a/prod_64.html

[toli]

I must say I'm not sure I trust the specs of the Chinese PS, I don't trust most thing coming from China

The Agilent is indeed better (BTW, I've measured the ripple in P-to-P value as well, and got 4.4mV which is about twice as much as the Agilent). Considering the one I have goes all the way to 40V (which is much higher than the Agilent - even though with only 1A), and has two channels, I'm more than happy with it, and I doubt I will need something better as far as a bench PS goes, and I don't deal with high voltages or currents (at least not at this time ).

[Kiriakos-GR]

I just tested my own PSU with my Agilent U1272A and I need some help. 

I set the PSU at 10.003V 
I started the Peak mode and I got as Max  10.008 and as min 10.002,
well its an ultra expensive old Kenwood , but toli is that of what you measured as micro volts  ? or millivolts ?
 

[toli]

Depending on the reading. I've mentioned in every place if its in V,mV, or uV. With the 10V (or the 8.4V for that matter) measurement the 1253A will only give me mV resolution.

[saturation]

Hi Toli,

I agree too, but I should have linked my own tests; given I can see AC + DC with the Agilent, I have a 1252a and a 1272a, why use the scope! Grrrr, so I retested my Mastech PSU and the AC component is 0.01% - 0.03% almost akin to the website post, from 1V-31V, @1A, however at 2A to 3A, it rises to a max of 0.3%.  I also confirmed this with the Rigol, no more than 400u VAC rms regardless of load or output voltage, all these in constant voltage mode.

https://www.eevblog.com/forum/index.php?topic=3171.msg42269#msg42269


These are just for comparison so you know where your specs are compared to a relatively recently made PSU, a top end PSU, and a 'reference' PSU.  All linear.  I think when you see the costs and the effort to get down to 150uV ripple on the lamdba, your device is still good by 2011 standards.

I must say I'm not sure I trust the specs of the Chinese PS, I don't trust most thing coming from China

The Agilent is indeed better (BTW, I've measured the ripple in P-to-P value as well, and got 4.4mV which is about twice as much as the Agilent). Considering the one I have goes all the way to 40V (which is much higher than the Agilent - even though with only 1A), and has two channels, I'm more than happy with it, and I doubt I will need something better as far as a bench PS goes, and I don't deal with high voltages or currents (at least not at this time ).

[toli]

Hi Saturation, my Agilent gives me a reading similar to the scope (a bit lower ripple than the scope actually, as I've mentioned earlier), but I don't know if I can trust it to measure AC voltages which are under a mV, as the error in these regions is relatively high. I've also measured my Sigma11 which I know has a much lower ripple, and the reading was about the same, and that means the Agilent isn't accurate enough at the bottom end of that range.

Other than that, I must agree that the performance of this unit isn't bad at all, especially when considering my needs, and the price I've paid for it.

I guess adding a better quality, and perhaps a larger value cap on the output (the one in there is getting old) will probably reduce ripple a bit more.

[saturation]

Hi toli,

Yes, you are right.  Always good to confirm with other devices.  You can confirm the AC reads by moving the DMM to mVAC, using other DMMs and non-DMM like a DSO to correlate the readings.  If they all say the same thing, the chances are very high the measure is true.

Hi Saturation, my Agilent gives me a reading similar to the scope (a bit lower ripple than the scope actually, as I've mentioned earlier), but I don't know if I can trust it to measure AC voltages which are under a mV, as the error in these regions is relatively high. I've also measured my Sigma11 which I know has a much lower ripple, and the reading was about the same, and that means the Agilent isn't accurate enough at the bottom end of that range.

Other than that, I must agree that the performance of this unit isn't bad at all, especially when considering my needs, and the price I've paid for it.

I guess adding a better quality, and perhaps a larger value cap on the output (the one in there is getting old) will probably reduce ripple a bit more.