Amrel PPS-2322 Programmable Power Supply Teardown

I picked up an Amrel PPS-2322 programmable power supply from eBay recently. You probably have not heard of the Amrel brand before as American Reliance (Amrel) no longer makes power supplies nowadays. Its power supply division was acquired by AMETEK Programmable Power, a subdivision of AMETEK back in 2010. While Amrel power supplies had been discontinued, Amrel branded electronic loads are still being actively developed and sold by AMETEK.

Amrel PPS-2322 is a dual 32V/2A programmable power supply. It has two fully independent channels. They can be used either separately in independent mode or used together in tracking mode. The power supply can be programmed via either the front panel keypad or via GPIB. Power supply outputs can be accessed via the front panel binding posts or the rear panel screw terminals. Note that the sense terminals are only present on the rear.

PPS2322_Front PPS2322_Back

The top case cover is fastened by the three screws located at the back of the power supply. Unlike the typical test instrument case construction, this case does not slide off the unit. To remove the cover you will need to slide it back a little bit, pry the bottom of the case outward and then lift the case upward. After the top cover is removed, you will see that it was actually constructed to have a slightly narrower opening at the bottom so that the tension makes the cover a very snug fit.

The picture below to the left shows the interior upon removing the case cover. As you can see the layout is rather clean. The main board at the bottom handles all the analog stuff and the plug-in processor board handles the programming and GPIB communication.

PPS2322_00 PPS2322_01

The processor board contains a CPLD (GAL16V8B) chip, two SRAM chips (MPM81C55P-2 is a 2K SRAM and SRM2016C is a 16K SRAM), and a 80C31 microprocessor (DMC60C31E-A, which is essentially a ROM-less 8051). The NI chip (CP19161) is presumably the GPIB controller (although I could not find any information on this chip). The firmware is stored in the UV-erasable EPROM (the firmware version of my unit is 5.23B).

Everything on the digital board is powered by a linear regulator 7805. The entire processor board is isolated from the remaining power supply circuit via an array of optocouplers.

PPS2322_03 PPS2322_04

Clearly, this 74HC373 was not seated properly before the board was soldered, another chip on the main board also has the same issue. I am guessing that the pick and place machine must had some difficulty properly inserting these chips. On the back of the board, there are couple of bodge wires. Other than that the layout is pretty clean.

PPS2322_02 PPS2322_21

The auxiliary power supply board is fitted on four plastic standoffs above the main board. The two large filter caps are 10,000µF 63V each, although they are only rated at 85°C. Since the caps are placed far away from the main heatsink, I suspect that temperature wasn’t really a concern. Looking at the bottom of the board It seems that the PCB for the auxiliary power was designed by hand.

PPS2322_05 PPS2322_06

The C-core transformer is an absolutely thing of beauty, it is located towards the front of the unit. The reason for the placement of the transformer is mainly for weight balancing since the heatsink is located towards the rear of the case. Though the transformer is much heavier than the heatsink so the case feels a bit front heavy.

PPS2322_07 PPS2322_09

From the Q.C. sticker, it seems that this transformer was made back in 1984. Although the date codes on the ICs and the silkscreen suggest that this power supply was make in 1995.

PPS2322_08

I did not bother taking off the front panel as there wasn’t anything interesting besides a two line character LCD and a keypad. On the left below is a picture of the backside of the LCD, it looks like a pretty standard 16×2 LCD. The picture on the right shows the socket on the backside of the keypad board, a couple of keypad controller ICs are also visible.

PPS2322_10 PPS2322_12

One thing interesting is how the wires to the front terminal posts are routed. If you follow the red and black wires from the front terminals you will see that they end up all the way at the back side of the power supply near the output terminal blocks. One issue to this layout is that because the wires are relatively long, the voltage drop across them can be quite significant. At full 2A load, the measured voltage drop approaches 10 mV at the terminals.

Since the front outputs do not have the sense terminal, you will need to use the rear terminal block connections in situations where higher precision is needed.

PPS2322_11 PPS2322_13

The two pictures below are the closeups the main board. In the left picture you can see one of the 74HC14 was not seated properly before soldered, similar to the problem we saw earlier on the processor board. And in the picture on the right, you can see all those 4N35 optocouplers similar to what you saw on the processor board. In a lab-grade programmable power supply, opto-isolation is critical as you do not want the power circuitry to destroy your entire system during a malfunction.

PPS2322_14 PPS2322_15

The picture below to the left is a shot of the processor board socket, you can see the date code (9509) on the edge of the PCB, which clearly indicates that it was manufactured in late 1995. On the other picture, you can see the DAC used in this PSU. It is an Analog Devices AD7541 12-Bit DAC. The output voltage resolution is determined by this DAC and at 4096 steps, the minimum adjustment interval for the output is just under 10mV.

PPS2322_16 PPS2322_17

In this portion of the main board, you can see three ICs with markings sanded off. I am not entirely sure what these three chips are used for here but clearly it must be something Amrel thought was important enough to put the effort in to obfuscate.

PPS2322_18 PPS2322_19

As mentioned earlier, the front panel binding posts are connected to the outputs all the way at the back of the unit, and here you can see the black and red wire running across the board.

PPS2322_20

Here is a picture of two heat sinked power transistors (TIP31C and TIP32C). All heat sinks in this power supply are securely soldered on via multiple points.

PPS2322_22

The pass transistors used in this power supply are 2SD1047‘s, two for each channel.

The main heatsinks (there are two, one for each channel) for the four 2SD1047s are rather small compared to the ones I saw in other power supplies. But they stay within the desired temperature even under full load thanks to the powerful cooling fan. Unfortunately, the fan is not temperature controlled and is always on. It is quite noisy.

I may add a fan controller to it (like I did with my Topward 6603A a few years ago) later.

PPS2322_23 PPS2322_24

The oscilloscope capture below are the power-on and over-current protection characteristics. The power on voltage curve is nice and clean and over-current cut off looks quite excellent as well.

PPS2322_TurnOn PPS2322_TurnOff

PPS-2322 also has over-voltage protection capability. Here is an oscilloscope capture of the OVP in action. The output voltage was set to 12V and the OVP level was set to 10V. The output was initially turned off and the waveform was captured when the output was switched on. As you can see when the voltage reached to around 10V, the output was automatically cutoff. The entire process took less than a couple of milliseconds.

PPS2322_OVP

One thing interesting about this power supply is that there is not a single relay inside. This means that there are no taps in transformer’s secondary to select from when the output voltage changes and thus the same input voltage is being bypassed via the power transistors at all times. So when the output voltage is low and when the current approaches the limit, each power transistor will dissipate more than 40W of power. But, as mentioned earlier the heatsink never gets really hot even under the maximum power dissipation.

At lower output voltages though, you do suffer the significantly decreased efficiency compared to other power supplies that select different windings depending on the outputs. But for a linear power supply, efficiency is not too much of a concern anyway.

Since there is no relay, the output on/off control relies entirely on the output transistor being turned on or off. In theory, the output on/off should be done using a relay since without a relay a connected device can inadvertently damage the power supply even when the output is off. But in practice, as long as you don’t try to connect a high voltage source to the output terminal when the output is off everything should be fine. In general though, you should never connect a low impedance voltage source to the output terminals of a power supply as most power supplies cannot sink current and doing so may damage the power supply.

Be Sociable, Share!

19 Comments

  1. Anton says:

    Great teardown, I look forward to reading these. I recently did one here (http://www.eevblog.com/forum/testgear/amrel-el1132-teardown/) on an AMREL load in a similar case. I found the engineering to be quite nice, but like yours there were a few minor build quality issues.

    The GPIB IC is a NAT7210. The name is right under the ‘National Instruments” logo, and kind of blends in with it.

  2. LJR says:

    Why do you think a programmable power supply is going to have a relay to switch ranges? It is designed to allow a voltage to be smoothly varied from 0 to max. A relay switchover midrange would introduce wicked transients. That’s perfectly acceptable for a manually controlled bench top PS but unacceptable for a programmable unit.

  3. LJR says:

    I just checked mine – it’s a 1326 and it does have two relays that switch between 16 and 32 volt ranges. The thing is, each range goes from 0 to either 16 or 32 and this avoids switching transients midrange. If I wish to switch ranges I have to do it programmatically or from the front panel.

    Your model doesn’t have the low range – which is too bad because the low range provides up to 6 amps of current.

    • kwong says:

      1326 is a one channel unit. so the relay is used to configure the secondary in either parallel or series.

      I suspect that they could get away without using relays on the secondary because the power dissipation is relatively low (at the peak each channel dissipates around 100W over two transistors). For linear regulators, the noise introduced due to the input winding switching is really not an issue if correctly designed (e.g. the input RC constant > than the switching time).

  4. Jules says:

    Hi Kerry; I was wondering if you could help me, i’ve looked all over the internet trying to find a rom image for my faulty 2CH PPS-2322 – would you be able to read the ROM contents into a .bin file?

  5. Alan says:

    Would you happen to have a schematic for the PPS-2322?

    • kwong says:

      No, unfortunately I couldn’t find it anywhere.

      • Alan says:

        Same results here. I wrote the company that bought out Amrel on a long shot and they were of no help. I discovered a couple of components that are bad. Cracked diode and a cooked resistor that looks like it is wearing a hula skirt. Luckily one of your pictures shows the resistor so I can use the color code from that. Thanks a bunch for taking all these pictures!

  6. Thanks for posting this information. I have used this wonderful power supply for 25 years for development. It is particularly useful because it resolves 10 uAmps of current.
    I have an issue with the keypad. Over the years it has become more erratic and I would like to clean it. After a quick look I cannot see how to access it. I have removed the power transformer chassis and 3 screws at the top of the main chassis behind the front panel. The top of the panel seems to be loose. There are also two similar screws at the bottom of the main chassis but I am reluctant to loosen the tie wraps on the wiring harness and they are very difficult to access. Will undoing these screws give me access to the back of the front panel and particularly the keypad?
    Thanks
    Dave

    • I guess I answered my own question. There are 5 screws securing the front assembly. Two are at the bottom and very difficult to access with the transformer in place. I undid the transformer mounting plate, 2 screws on each side and two underneath. The transformer can then be folded back for easier access to the bottom screws of the faceplate. There are two tie-wraps holding the wiring and preventing moving the transformer assembly far enough. I put two new tie-wraps around these and pulled them loosely. Then I cut the original ties. In this way, I can tighten them up as the original when I’m finished.
      The front plate can then be released using a long (very long) phillips screwdriver. The pushbuttons are secured with 4 small screws.
      I experimented with several methods of cleaning the pushbutton assembly. I cleaned the circuit board and buttons with alcohol. I again assembled the power supply, but there was no improvement. I tried three more experiments, each time reassembling the supply. Finally, I cleaned the circuit board with Radio Shack ‘Contact and Head Cleaner’ Nexxtech 4408726 but any contact cleaner should do. Then I sanded the carbon area on the pushbuttons VERY lightly with 1000 grit sandpaper. (i.e. one wipe. There is apparently enough carbon on the buttons to do this. This did the job and the buttons respond like new.

      PS. The power supply actually resolves 100 microamps of currrent. Very few newer power supplies do!

      Thanks for this forum.

      Dave

  7. manmohan krishan says:

    hello averyone
    sir i have a amrel power supply model pps 1326 .it is 2 output range 16v6amp or 32v3amp.
    its 16v6amp range work proper but 32v3amp range not select.
    please help me how i can select this range.

Leave a Reply to David Rathgeber