EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: TheUnnamedNewbie on August 04, 2020, 10:27:38 am
-
My place of work does a lot of IC work. As a result, in the first verification stage, we need a lot of bias and supply voltages. Setups with 10-20 supplies or more are not unheard of.
Right now this is often done with a combination of lab supplies and custom-made bias boards. Bias boards often use trimmer resistors/LDOs to generate the voltages.
I've been looking at ways to do this entirely with programmable lab equipment. The problem I keep running into is that there is no 'ideal' solution for us. Lab supplies often are bulky (when you consider size-vs-number-of-outputs). They also require lots of different instruments (if you want 20 supply voltages with 2-channel supplies you need to keep track of 10 instruments), which opens you up for a lot of programming errors when setting up everything for automated scripts. There are PXI modules by Keysight, NI and others that offer a bunch of analog outputs (eg, 32 16-bit voltage outputs) which are ideal for the bias voltages in our chips, but just lack the current capability we need for supplies (but only just - most of them are 10 or 20 mA capable, but we often need 100 mA on a handful of supply rails).
Does anyone know of a nice hybrid? Something that has more than the 3-4 channels you find on lab supplies, can generate precise voltages (at least 1 mV step), and can output say 100 mA?
Right now the only option I can think of is either using a PXI rack with 10 PSU modules, or using one of those analog sources, and just making a custom PCB with an array of unity-gain followers.
Keep in mind that 90% of our supply voltages are less than 1.2 V.
-
Have you looked at these: https://www.ni.com/en-us/shop/hardware/products/pxi-programmable-power-supply.html#:~:text=PXI%20Programmable%20Power%20Supply%20%20%20%20Model,to%2010%20V%20%201%20more%20rows%20 (https://www.ni.com/en-us/shop/hardware/products/pxi-programmable-power-supply.html#:~:text=PXI%20Programmable%20Power%20Supply%20%20%20%20Model,to%2010%20V%20%201%20more%20rows%20)
-
Yeah, I've come across those, but felt like they still lacked the output channels, and are expensive for what they offer. The only advantage they have over dedicated supplies is that they would fit with multiple in a PXI chassis. But they seem to be well over twice as expensive as similar performing dedicated supplies from keysight or R&S.
Right now, I'm keeping them as a kind of back-up option: I could consider setting up PXI racks with maybe 2 or 3 of those PSUs combined with a 32 channel analog output card.
-
It doesn't meet your current requirements, but you can mount 10 dual channel DACs on some models of LabJack.
https://labjack.com/support/datasheets/accessories/ljtick-dac
Yes, you'd still need to add the unity-gain followers to boost currents as they are pretty limited. Otherwise a flexible but potentially messy setup.
-
Something like a Agilent 4141B (Semiconductor parameter analyzer) packs a lot of things used to test semiconductors into one box. It has a 4 SMU channels with up to +/- 100V at up to 100mA (CV or CC with resolutions down to picoamps) along with 4 voltage ports that just output or sense voltage. Its pretty overkill but can be found cheap on the used market. Tho you still need quite a few for that many channels.
You could also design something like this yourself. Just grab a microcontroller, give it a USB to UART chip for communication to the PC, then stick on bunch of I2C interface DACs followed by high current opamp buffers. The low voltages involved make for a lot less heat dissipation in those high power opamps so you could proabobly stuff 32 channels of this into something the size of a VHS cassette box.
-
It's a hell of a price premium for power supplies, but PXI SMU modules can often source or sink 100mA or more. Something like the PXIe-4141 will do 4 channels, +-10V, 100mA a channel. Also takes care of precision measurement of the draw, if you need it, and can be mixed with higher voltage or current modules in the same chassis and driven by the same software.
If you're willing to prototype a board for it yourself, you could also build something that takes the outputs from one of those many-output analog cards and drives a bunch of power BJTs, using the outputs as the voltage reference. Maybe a few traps in there with temperature compensation or if you need the loads measured, but if you just need a lot of outputs, it could be a less expensive option.
-
You could also design something like this yourself.
I want to avoid designing things myself as much as possible, especially if it has to be programmable. We have already had a few attempts of people doing this, but it always gets dropped because there is no commercial support for it*, and people don't want to bother figuring out how it is supposed to work (keep in mind a lot of these people are analog/RF engineers, and aren't the best embedded programmers - automating our measurements is already a big learning experience for many when it comes to programming and scripting).
The PXI SMU units do look interesting. It is unfortunate that it costs so much because of the 4-quandant nature, but it might be worth it. Especially since the measurement is built-in. If only people could do that without the 4-quadrant part, and just make it single-quadrant, that would be great.
Currently, it is between that, or making a simple analog buffer board (which has no programming, so it's acceptable I think) that allows people to buffer a number of 'high-current' (ie, >10 mA) lines and using a simple analog-voltage-source PXI module.
*What I mean by this is that person A spends 2 months designing their big multi-function board with lots of features, with the intent of using it themselves but then offering it to others. After their measurements, maybe their office-mate uses it too, asking a bunch of questions to the designer. The designer then leaves, and when a third guy wants to use it, they notice a lot of half-done or inconsistent documentation, decide it is too much work, and then just give up, and either design their own or go back to making a board with a bunch of trimmer pots instead.
-
Choked on my coffee when I saw the GBP8500 price on those PXIs.
There are much better prices on ebay.com
-
Choked on my coffee when I saw the GBP8500 price on those PXIs.
There are much better prices on ebay.com
May very well be, but for a commercial lab application, like mine, the hassle of having to deal with ebay, and not getting support, etc, is not worth the few k we save. You would probably need to have your ebay modules calibrated again before you can really rely on them, do a performance verification, etc.
Also keep in mind that when buying a number of modules or devices like this, nobody pays list price. You get discounts, making the money saved by going to ebay lower.
-
What kind of voltage range do you need? There are a few lower cost PXI options available from Pickering.. 2 10V supplies in a single module with 16-bit set point: https://www.pickeringtest.com/en-us/product/dual-programmable-+10v-power-supply (https://www.pickeringtest.com/en-us/product/dual-programmable-+10v-power-supply) Or, a battery simulator may be a workable option if the voltage range is workable: https://www.pickeringtest.com/en-us/product/6-channel-battery-simulator-module (https://www.pickeringtest.com/en-us/product/6-channel-battery-simulator-module)
-
Have you considered looking at older technology?
For example a bank of Tektronix PS5004 programmable power supplies mounted (6 per 19 inch rack) in a TM5006 chassis would appear to suit your needs and consumes less space than many other options. These power supplies will provide 0 to 20V in 0.5mV steps at a current of up to 300mA with 4W sensing for accuracy. Control is by manual front panel controls or via GPIB.
These are now discontinued but are still available from time to time on various auction sites. I believe these were also made under the Tegam brand after Tektronix discontinued the product.
-
Many waveform generators will do this, just set it to 0Hz, 0V amplitude, and the offset being your DC voltage. 20mA should be no problem. (The Siglent SDG2042X even has a DC waveform.)
-
In the first post OP says he wants 100mA output current.
I think that rules out all sig gens?
-
I would trawl eBay for HP6624A and HP6626A, four outputs each.
-
Wouldn't 5 4 channel power supplies suffice? It's a modest 1 metre shelf of kit and nothing too fancy or unusual. Easily used elsewhere too.
-
In the first post OP says he wants 100mA output current.
I think that rules out all sig gens?
Depends on the device. As an example, the SDG2042X specs ±200mA output. But of course it has a 50Ω source impedance, so would experience a voltage drop. But on the plus size, it's a fixed 50Ω source impedance so could be adjusted for.
-
Since he only needs about 1V you could use a 1:10 divider to drop the voltage and bring the output impedance to 5Ohm. But its rare to find high channel count signal generators, and they tend to be quite a bit more expensive that power supplies.
But yeah if you already have an 32 channel analog output PXI card then id just use a bunch of high power opamps to buffer that. Its a fully analog solution with no software involved that can easily be reused elsewhere to buffer any other voltage you like. Doesn't even need to be built on a proper PCB, can just take a bunch of TO220 opamps stuck to a heatsink and hand wire them into voltage folower mode. Easily done in one afternoon once you get the parts from DigiKey or whoever you buy from.