I bought two of these very small
Keysight E36103A PSU's from our fellow forum member Faith and my curiosity required me to open one up and see the inside of Keysights small PSU series.
Here are some pictures of the tear down.
Some measurements will follow.
More pictures from the inside.
Mostly noticeable is the relative huge toridal transformer.
It is interesting to note, that all pictures I found online from Keysight have the screw type binding post to be used with spade connectors.
But the one I got hat low cost -only plug in- 4 mm banana plug connectors.
- Picture 1 is from Keysight
- Picture 2 is from my E36103A
No fan noise, even at 2A, the PSU is very quiet.
The display is easy to read from a good distance and the interface is much better, compared to older Agilent PSU's
Nice snatch. I believe seller mentioned that shrouded terminals are an option for smaller depth
.
Bleed resistor in place of capacitor looks funny. Moar photos..
Unpopulated PLCC and two footprints nearby STM MCU are for GPIB.
Nice snatch. I believe seller mentioned that shrouded terminals are an option for smaller depth .
Bleed resistor in place of capacitor looks funny. Moar photos..
Unpopulated PLCC and two footprints nearby STM MCU are for GPIB.
Yes, the user guide shows option J01 as "recessed binding post" configuration.
May be this option cost even more!
The power inductor looks really cheap
Spot welding of the housing looks not good, at least not Keysight quality
Values are 0.02 Ohm and 382 uH
This inductor is only half way potted in its housing, hmm....
Interesting, your observation for GPIB option.
May be this PCB is used for other PSU's as well, the housing has no option for a GPIB connector.
Some more pictures of the I/O board with USB and LAN connector
Here is the connection of the inductor to the power board
The inductor is connected to J101
On the bottom board we see a C101 missing, hmmm
And here is the output to the binding posts
The LXI web interface works really well and right away with the "Keysight" password.
Taken control from this interface, everything works, all knobs are active.
The response is fast and all seems workable.
But, taking a screenshot is a joke.
There is this nice "
screenshoot" (Spelling mistake by Keysight on purpose?) button
But it saves the screen as a .BMP file, so small you need a microscope to look at it.
Why?
But it saves the screen as a .BMP file, so small you need a microscope to look at it.
Why?
Most probably because they just copy the data from the front panel display to the bitmap.
And since the front panel display has a resolution of 128x64 pixel, the bitmap file has the same size.
Scaling the data could be difficult because the fonts are stored as bitmap data, not as a vector font.
One of my two E36103A had failed a few weeks ago, showing some random numbers of current, even if nothing was hooked up.
Then I sent it to Keysight for repair under warranty and received a brand new E36103A in exchange.
A big thank you to Keysight for honoring a warranty claim and making a fast exchange.
The new unit came with a new calibration certificate.
But two things are very interesting:
1. My old unit had isolation banana plug receptors and the new one has the nice binding posts
2. Just funny: They placed the new CAL sticker partially right on the display protection film.
Beside this glitch of one of the two PSUs, I am really happy with them and can highly recommend this Keysight PSU.
Here are some pictures of before and after the exchange.
I like to buy Keysight, because I do not only get great value but also great customer support, as in this case with an exchange. In addition all my Keysight instruments are nicely integrated with BenchVue. On top of that, I did not pay a full price for the two E36103A. And lastly, if I want to sell a Keysight instrument down the road, I am getting usually a real good amount of money, even if the instrument is may be 10 years old.
Well, back on topic, I just did a few tests with my replacement E36103A and it is just perfectly in specs.
Here I measured the current stability at 2 A, Only 1.25mA Pk/Pk over 2600 readings in 4 wire sense mode.
Here are some measurements with a current probe, turn ON and OFF
PSU set to 2.000 A CC
You get your upgrade to proper binding posts, kudos to HPAK.
Thanks for the review, this Keysight PSU is nicely put together and compact but, for a price of over $1,000 (that's what I saw on on eBay and Amazon) there are a lot of other supplies that would seem to me to offer much better bang for your buck, multiple outputs, bigger displays. Now if it were going for $500 then we might be in more reasonable price territory.
You get your upgrade to proper binding posts, kudos to HPAK.
Now, that I have one of each, I prefer the PSU with the proper screw on binding posts.
Well, Keysight is just amazing, when it comes to customer service!
I had both of the E36103A PSUs running over night at 1A and 2A CC and they are just very stable.
Thanks for the review, this Keysight PSU is nicely put together and compact but, for a price of over $1,000 (that's what I saw on on eBay and Amazon) there are a lot of other supplies that would seem to me to offer much better bang for your buck, multiple outputs, bigger displays.
Not if you want better accuracy and low current measurement capability. With a cheaper power supply you'll need to spend extra money if you want to measure the current into a load accurately without causing too much voltage drop. If you are into designing low power (microcontroller) circuits then these or the Keysight E36312A / E36313A are the power supplies to go for.
The 3610x series starts to look kinda expensive if you compare them with the just released 3631x series.
They have multiple outputs, with possible parallels and series connections, low current measurement, and a big color display.
I hope that Keysight would consider dropping the price a bit.
If they made them at around $500 I may grab one.
But again, the 36312 at 1200 euros (add 20% VAT) with all its features, look a much better deal.
Not if you want better accuracy and low current measurement capability. With a cheaper power supply you'll need to spend extra money if you want to measure the current into a load accurately without causing too much voltage drop. If you are into designing low power (microcontroller) circuits then these or the Keysight E36312A / E36313A are the power supplies to go for.
If 4 wire voltage sensing at point of delivery and uA current displays are an essential requirement then I suppose I might part with $1,700 for a E36313A but, as a business owner, I have to be realistic about what I can afford.
I do designs for low current - one of my latest projects is pulling nA at the lowest power settings - a few uA at most so for this, I used a $350 Siglent SPD3303X-E power supply in series with a Keysight 34461A meter; at such low currents, the voltage burden of the current meter in series is insignificant and the Siglent power supply seems to be comparable in terms of accuracy to the Keysight supplies - it certainly works for my needs.
I'm not trying to be contrarian, just pointing out that there are opportunities to meet a requirement and save money when funds are not unlimited and, as you can see, I'm not against paying for good stuff when needed, especially when you consider the awesome level of support that Keysight provides. I think Highvoltage has already stated that he didn't pay list for his PSUs.
Not if you want better accuracy and low current measurement capability. With a cheaper power supply you'll need to spend extra money if you want to measure the current into a load accurately without causing too much voltage drop. If you are into designing low power (microcontroller) circuits then these or the Keysight E36312A / E36313A are the power supplies to go for.
If 4 wire voltage sensing at point of delivery and uA current displays are an essential requirement then I suppose I might part with $1,700 for a E36313A but, as a business owner, I have to be realistic about what I can afford.
I do designs for low current - one of my latest projects is pulling nA at the lowest power settings - a few uA at most so for this, I used a $350 Siglent SPD3303X-E power supply in series with a Keysight 34461A meter; at such low currents, the voltage burden of the current meter in series is insignificant
I wouldn't call a 100 Ohm current sense resistor in series with a power supply insignificant. Many microcontroller/SoC circuits go from several uA to tens of mA (or more) when going from sleep to normal operation. That won't work with 100 Ohm in series because the DMM will need time to switch ranges. This is not considering current surges (Wifi and 'GSM' modems are notorious).
If 4 wire voltage sensing at point of delivery and uA current displays are an essential requirement then I suppose I might part with $1,700 for a E36313A but, as a business owner, I have to be realistic about what I can afford.
My TTi QPX1200SP will do 1.2KW with powerflex and can be set in mV and 10mA. If I wanted smaller current mesurements I'd use another measuring device. They also support remote sensing. all for £/$ 1150
nctnico All current meters have a burden voltage, the most expensive one I own, the $1,000 Keysight 34461A, has the following burdens on DC current, for the lowest 100uA range (with 1 nA resolution) which is a resistance of around 100 ohms. What's your point?
I suppose that it's useful if a power supply can read the supplied current accurately and still maintain output voltage but how many power supplies have nA resolution/accuracy on their output metering? With 4 wire sensing, you could put the sense on the load and the supply would regulate to take account of the current meter's burden voltage, I can do that today with an old HP PSU that has remote sensing.
If 4 wire voltage sensing at point of delivery and uA current displays are an essential requirement then I suppose I might part with $1,700 for a E36313A but, as a business owner, I have to be realistic about what I can afford.
My TTi QPX1200SP will do 1.2KW with powerflex and can be set in mV and 10mA. If I wanted smaller current mesurements I'd use another measuring device. They also support remote sensing. all for £/$ 1150
That PSU is a beast! It's sold at tequipment.net for $1,500 and you could get a 6% discount with EEVBLOG.
Of late, I am working on designs that are supposed to run for a year on a 225 mAhr CR2032 battery and I often use an actual battery as my supply beacause the battery itself has significant internal resistance. I designed a 225 kW PSU a while back that had a programmable internal resistance so it could act as a Hybrid vehicle battery simulator.
My point all along has been that selecting the right gear is a "horses for courses" exercise that's often constrained by that fact that you have to work within tight budgets and so there are many ways to get the accuracy you need.
nctnico All current meters have a burden voltage, the most expensive one I own, the $1,000 Keysight 34461A, has the following burdens on DC current, for the lowest 100uA range (with 1 nA resolution) which is a resistance of around 100 ohms. What's your point?
I suppose that it's useful if a power supply can read the supplied current accurately and still maintain output voltage but how many power supplies have nA resolution/accuracy on their output metering? With 4 wire sensing, you could put the sense on the load and the supply would regulate to take account of the current meter's burden voltage, I can do that today with an old HP PSU that has remote sensing.
I'm quite sure that won't work. The sense circuit is designed to compensate for losses in wires which have a low resistance (less than 1 Ohm). If you put a 100 Ohm resistor in series with the output and try to compensate the voltage drop across this resistor you'll get a slow response because the output capacitors need to be charged & discharged. I wouldn't be surprised if you get loop stability problems as well. Dave has designed his uCurrent because DMMs are not the right tool for measuring low currents into dynamic loads like microcontroller circuits.
The voltage between the force and sense terminals is usually limited to something like 0.5 - 1 V (check the manual of that particular power supply). So a 100 Ohm resistor might work if you are sure the peak power draw is below 5 mA. Obviously this would tie up a DMM for the duration of the measurement.
The voltage between the force and sense terminals is usually limited to something like 0.5 - 1 V (check the manual of that particular power supply). So a 100 Ohm resistor might work if you are sure the peak power draw is below 5 mA. Obviously this would tie up a DMM for the duration of the measurement.
Didn't you read my previous posting? Think about what happens with the output capacitors in the power supply and you'll see it won't work with such high resistances in series with the load! The E36103A is rated for 2V max. compensation (1V in each lead). Because this is a 2A PSU this translates to a maximum (total) resistance in series with the load of 1 Ohm.
If they wanted to limit the resistance between force and sense, they would have specified a maximum resistance. But they specified a maximum voltage. As long as your I*R drop does not exceed this voltage, you are within their specifications. As for the output cap, I am pretty sure these supplies have an active down programmer. Without that, I agree that transient response will be degraded. How much output capacitance do these power supplies have? I expect output capacitance to be quite low in a modern design.
For best transient response, Keysight
suggests (page 15) a capacitor at the device under test. Obviously that would degrade the current measurement bandwidth.
This document discusses this issue in some detail (including a discussion of bypassing at the DUT).
Decoupling is tricky because most typical decoupling capacitors leak. Electrolytics can leak up to tens of uA. Ceramics are in the nA range.
Linear power supply with constant current capabilities must have a current sense resistor anyway for the purpose of current regulation.
And they of course take great care during design so that sense resistor have minimum effect on the regulated output.
So why not do the thing from the ground up, and have precision current measuring from it, instead of relying on external measuring devices, which is more error prone and requires care to obtain precise measuring.
Of course, this is may be more costly. And I fully understand that people having a PSU without precision current measurement but a 6.5 or 5.5 digit bench DMM, don't want to buy an extra PSU, and would prefer to make the most of the (costly) equipment they already own.
Assuming you do not need said DMM for something else. If you do, for example to measure something else in your device under test, then buying a power supply that can do the measurement on its own rather than than buying a cheaper power supply + another DMM may make sense.