Products > Test Equipment
TTi MX100TP triple power supply teardown and review
MarkL:
I recently purchased a TTi MX100TP triple output power supply and wanted to share some info and pictures. I got mine from tequipment.com for US$1280 minus the eevblog discount. I've only had the unit for about a week now, so it hasn't seen much bench action.
I bought it because my go-to supply, an Agilent E3631A, sometimes didn't have enough current output to suit the task at hand, and the quirky user interface was starting to drive me bananas.
There are two models of MX100: the MX100T and MX100TP. Both have 3 fully independent and isolated outputs with a total of 315W. The "P" version has added remote programming with RS232, Ethernet, GPIB, and USB and is US$240 more than the non-P version. The official spec sheet is available via the product page; I won't repeat it here. I never had anything else from TTi except their iProber, but I was impressed enough by the MX100 specs to try it.
Here's the product page:
http://www.tti-test.com/go/mx/index.htm
Part 1 - Some mechanical comments.
The MX100TP is very light compared to the E3631A, but it's almost exactly the same size.
The front panel is simply laid out with a single button to setup each channel, and to turn each one on or off. There's also all-on or all-off. The big encoder knob does not push for "ok".
The back includes terminal strips with connections in parallel to the front panel for rack use, but I don't think I saw a rack-mount kit.
The fan pulls air in through slots in the front panel and also from open holes underneath. The slots in the front are stuffed with air filter material but the bottom is open. Go figure. The fan is variable speed and is, in my opinion, very quiet.
Binding posts are decent quality with metal inserts and operate smoothly, but they only open 1/8". They have standard 3/4" spacing.
Inside there's lots of empty space. I think they could have spent some time to make the the unit a little more compact. The back and bottom are steel, with two thin aluminum rails to support an all-plastic front panel. With the cover off, it's very floppy. The use of plastic speed rivets give it a cheap look. For the price I was expecting better mechanical construction. All the electrical components are recognizable name brands: Pulse, Kemet, Schaffner, NXP, Elna, to name a few.
The design appears to be a two-stage switcher with a final linear output regulator. Here you can see the output #2 (op2) section. There's two heatsinks. One is a diode which appears to be rectifying a PWM waveform. The PWM is based on the output requirements. The other heatsink has a MOSFET. There are no switching waveforms on the MOSFET which is why I think it's in linear mode. It appears the switcher is acting as a pre-regulator to minimize the heat dissipation in the linear stage. The heatsinks get warm but not too hot to touch even under worst-case load.
Each output has at least two different voltage and current configurations with roughly the same wattage. In some combinations when higher output is selected, the unit will disable one of the other outputs and you only get a total of two outputs. (You will get a very clear indication if this is going to happen.) The outputs can also be configured to track each other (voltage only) in a few different combinations. Output #3 (op3) can have a higher voltage than the other two and has a beefier output MOSFET:
There's one big switching transformer that serves to supply and isolate each section. To its immediate left is a small switcher that supplies the front panel.
The control processors for each of the sections appears to be on the PCB right behind the front panel. There's another board behind it where the display and keypad are mounted. It may be hard to see, but the well-insulated AC power switch is in the extreme lower right and is brought to and from the front panel on wires and not on the PCB. The soldering on these and all the boards are good quality with minimal flux residue.
Everyone seems to be crazy about grounds. So here it is. Direct from the IEC socket, through a ferrite a couple times, and then to a crimped ring. Looks fine to me.
Each output has separate sense inputs. The sense input is enabled with a slide switch. I don't like slide switches at all since they get flaky. I thought putting a slide switch on the sense input was the worst idea I'd ever seen, since if the switch gets intermittent your voltage could fly out of control from the lack of feedback. But TTi seems to have designed it in such a way that the sense input has a very limited range that it can affect the output voltage. So, not having a sense input, or a flaky input, is not a danger to anything connected to the output.
But, I still need to complain about the sense input terminals. They're quite hard to get to, and they're deep so you need at least 1/2" of bare wire to get to the contact.
Another thing I noticed was that the GPIB jack screws aren't quite deep enough. So, the connector does not seat all the way. It seems to be far enough to make contact, but connectors shouldn't be loose.
(continued in part 2)
MarkL:
Part 2 - So, enough with mechanical stuff. How well does it work?
There's lots of output combinations (read the spec sheet in the previous post), but I thought I would set up the highest current and highest voltage configurations. Here's the max current config driving an Array DC load. The MX100 has its outputs paralleled and is outputting 16V @ 15A. You can see by the voltage reading on the Array there's a significant loss in the test leads at this current. I could have connected all three sense lines to compensate for this, but I didn't do it in this picture. (The sense inputs work fine; I did try them.)
Here's the outputs reconfigured for max voltage 140V @ 1.5A with the three outputs in series:
I found the readback voltage to be very accurate on all 3 outputs. On output #1 it was no worse than a few mV, and most of the time much less than that. Current was similarly accurate to a couple of mA. Output #1 has more precision in setting and readback than the other two, for some reason.
One thing I would have liked to see on the home screen is the watts being supplied by each output. There appears to be enough space for it. But you can still get the reading ("VxA") by pushing the configuration button for an individual channel:
Noise is well within spec, as far as I can determine. It's so low I had to pull out an old Tek AM502 differential amplifier. The limit on this amplifier is 1MHz and not 20MHz, which is how power supplies are usually specified, so take it for what it's worth. Here's the MX100 (output #1) compared to the E3631A (6V output). The amplifier is set to x1000, so these captures are in mV (not volts). Outputs were set to 5V into a 10ohm resistor. The DC load was not used since those tend to inject noise. The MX100 internal switcher runs at 100kHz and I was not able to see that frequency in the time domain but it does show up on an FFT (FFT not shown).
Here is the E3631A:
And the MX100:
A quick load response test stayed within a couple hundred mV. Using the DC load to create a 5A 1ms load pulse, here is the output voltage in yellow, current in green. Output voltage was set to 5V.
There was some glitchiness when changing ranges on the various outputs. All the outputs had this artifact. It wasn't extreme, and was always around 125mV. Here's an example. The output is loaded with a 10ohm resistor and is configured off.
The unit has a hard AC power switch. Power off could be a little better. Here is turning the AC off with open terminals.
And here it is again, this time with a 10ohm load.
And here is turning the AC power on (the output is configured off), again with a 10ohm resistor.
So, AC power on/off transitions could be handled a little more gracefully, but glitches of a few 100mV isn't likely to hurt anything.
I tried out all the communications interfaces. I use Linux and they all work with Linux. (I'm sure they work with windows too, but don't ask me.) The USB interface appears as an ACM device. RS232 is 3-wire xon/xoff. All remote interfaces are enabled at all times, although there are exclusive lock commands if your application needs it.
It should go without saying, but all interfaces are isolated.
One non-obvious thing is that all commands submitted through USB or RS232 must be terminated with 0x0a. Otherwise the unit will not respond. This is buried in the documentation but I missed it.
Any time a remote commands is submitted, it puts the front panel into local lockout. This is common behavior for GPIB devices, but I wish it could be configured. The alternative is to always send a command to re-enable the front panel.
Firmware upgrades are handled via USB. There aren't any upgrades yet that I am aware of. There are instructions on how to do it from Linux if you don't use windows. Service and calibration documentation will be available, but customer support says it's not done yet.
The unit will respond to commands submitted over a TCP socket, but it also has a web server. The web server will allow you to do basic configuration and status, but in terms of working with the instrument this could use a major amount of work. The only control possible from the web page is to submit text commands using this page:
They need to add a *real* GUI with buttons, knobs, real-time plots, and everything that you'd expect from a web interface now-a-days. If I wanted text commands I can already get that from telnet.
The PDF user manual is very well written. There's no sign that this is from another OEM or is a rebadge, but if someone recognizes it please post.
Comes with a printed manual and CD. CD contains the docs and drivers for all TTi products. Drivers: IVI, Labview, LXI, NI, USB, TCP example program. (All useless to me and only adds to my cost.)
I've only had the opportunity to use TTi tech support three times. Once about my iProber, and twice on the MX100 about the calibration/service manual and then about firmware upgrades. The questions were relatively easy, but answers were prompt and complete. So far, so good.
What I like most about this supply is its wide range of voltage and current outputs to fit just about any project. We'll see. I'll post any interesting updates, good or bad, as it starts getting used for real.
Thomas:
Very well written teardown and review! :-+
Looks like a very nice power supply with an intuitive user interface (important!).
This unit seems to fall in line with my impression of TTi: Solid performance and nice set of features for a reasonable amount of money.
My EL302RT triple power supply is probably designed on those principles too, it is a solid performer but could have been made smaller.
I am pretty sure this is not a rebadge or OEM, designed and built in England. Mine has "Made in England" on the back :)
BloodyCactus:
so glad you posted this. nearly bought one last week. I need to spend some time tonight and read over more what you wrote. 1.2k makes one think a bit before dropping cash.
MarkL:
--- Quote from: Thomas on April 10, 2015, 05:50:16 pm ---Very well written teardown and review! :-+
Looks like a very nice power supply with an intuitive user interface (important!).
This unit seems to fall in line with my impression of TTi: Solid performance and nice set of features for a reasonable amount of money.
My EL302RT triple power supply is probably designed on those principles too, it is a solid performer but could have been made smaller.
I am pretty sure this is not a rebadge or OEM, designed and built in England. Mine has "Made in England" on the back :)
--- End quote ---
Thanks!
Yep - I missed that. I checked and the MX100 also has "Made in England" in small print on the TTi sticker on the back. Nice.
Navigation
[0] Message Index
[#] Next page
Go to full version