MHS-5200A function generator teardown / review / reverse engineering

[HighVoltage]

Boy is that square wave ugly.....   With the new re-designed PCB...

It looks like this generator is more for simple hobby applications.

[bianchifan]

A couple of days ago I decided for a flash deal at Gearbest's, 51.74$ seemed pretty good to me
Obviously they are working permanently on the PCB and I received a fresh new version, although only minor changes.
Maybe manually reworked too, lots of soldering honey in the filter area and numerous soldering spatters.
Unfortunately the CD is mispressed, so no arbitrary for me at the moment, at MHINSTEK I found only chinese program version (.









Overshoot's still present, ringing cannot tell due to noisy scope...

All waves @1MHz

[EU1]

Given that these units are designed to a very low price point, my guess is that they somehow/somewhere got a really good deal on a large lot of AD812's.

AD812, AD603, LMH6612, polymer capacitors, dual coil inductors, relays, and probably flash are reused old parts.

But this is actually not too bad considering the price. If they used new parts, it wouldn't be so cheap ($56 including shipping in my case).

[EU1]

Unfortunately the CD is mispressed, so no arbitrary for me at the moment, at MHINSTEK I found only chinese program version (.

I've uploaded the software by the following link:
https://www.dropbox.com/s/zjdbeem2rsq5ngq/MHS-5200A.7z?dl=0

[technogeeky]

I, too, have the newer version. I paid about $90 for it because I needed it immediately to run some tests, and had planned to return it and buy the $50 version.

I will post some detailed scope measurements of the signal in a bit, but in the meantime, I figured I'd make use of my shiny new USB microscope and post pictures of all the critical components of this board, so you guys can get a better idea of what's inside. I can post pictures of the overall circuit board later (or you can use the ones above, which are apparently identical to mine). But none of the previous photographs have quality IC pictures, so I figured I'd contribute.

Although uploading 25 images on this forum is kind of tedious...

[technogeeky]

I continue in this post...

There are "missing" pictures above, but they were just deleted bad-quality duplicates. Fear not, no parts are missing.

edit: Evidently my rotation settings didn't save. Sigh. I give up.

[technogeeky]

Measurements of the newer-style MHS-5200A.

Test Conditions:

• both devices reset to default settings
• connected directly using 50-ohm (verified) BNC to BNC cable
• Rigol 1054z with 1x setting

Test 1:

• Control: 2.100 kHz (a reasonable minimum)
• Variable: 1, 5, 10, 15, 20 Volts
• Measuring: Rise Time, Overshoot, Vpp
  

[technogeeky]

Measurements of the newer-style MHS-5200A.

Test Conditions:

• both devices reset to default settings
• connected directly using 50-ohm (verified) BNC to BNC cable
• Rigol 1054z with 1x setting

Test 2:

• Control: 1 Volt peak-to-peak
• Variable: 40, 50, 60, 70, 80, 90, 100, 170, 180 kHz then 2.0 MHz
• Measuring: Fall time, picture
  

Bonus:

Whatever is going on here, animated.

[technogeeky]

Measurements of the newer-style MHS-5200A.

Test Conditions:

• both devices reset to default settings
• connected directly using 50-ohm (verified) BNC to BNC cable
• Rigol 1054z with 1x setting

Test 3:

• Control: 6.0 MHz (the fastest frequency which doesn't totally look like a sine wave)
• Variable: 1, 5, 10, 15, 20 Volts
• Measuring: Rise Time, Vpp, Overshoot

[Dwaine]

I wish we had the firmware to play with.   Also anyone probe the pads on the circuit board?   On the side. I like the unit. I just wish it had better performance.

[Glapsvin]

I just bought new one for 50$, as previously reported it does seem to be improved from the original terdown. I know the limitations of my new toy and so far  I'm very happy with the purchase (performance / price point).

[bianchifan]

I've uploaded the software by the following link:
https://www.dropbox.com/s/zjdbeem2rsq5ngq/MHS-5200A.7z?dl=0

thx, I found it in the meantime on MHInstek website.
Installed and running fine so long.
I wonder about the efforts with excel sheets and export stuff from alternative Software, I have nothing to complain at the moment.

GB did a refund about 25$ plus 125 extra points(2$50).. so no bad deal.   

[devttys0]

For those interested, I finally got a hold of one of the new MHS5200A units and did a short overview of the new features, plus a teardown, circuit analysis, and performance review.



TL;DW: The analog front-end underwent a not-insignificant redesign. The newer models generate much nicer square waves, at the expense of increased high frequency noise. Slew rate distortion is still an issue. Generally poor performance when compared to a "proper" signal generator, but for the price it's still a good bang-for-the-buck, especially if you're only interested in lower frequency signal generation and don't mind the HF noise in the output signal.

[nowlan]

I watched that video last night. Prolly good enough for my needs.

I missed the part whether this can do a sweep or adjust frequency via dial on the fly, or if needed to be paused/stopped to adjust.

[Dwaine]

Thanks for the video.   I'll try the 100pf cap....

[PoV]

Ah crap! I just bought one of these (newer 12bit model), but didn't realize I was buying the 6 MHz version instead of the 25 MHz version (was $1 cheaper). Doh! 

According to this manual I stumbled across, the only difference between models is the Sine way frequencies (all other generators seem capped at 6 MHz). Has anyone looked in to dumping/reprogramming the firmware yet? I would imagine they're exactly the same, just different firmwares.

[BugCatcher]

I bought a MHS-5225A release R5.04
started to play with and found following "challenges":

1) Frequency Error
    - connect "In ext." with "ch 1"
    - set frequency to 20.000 000 00 MHz
    - run the internal frequency counter with 10s gate time
    -> my result is 20.000 146 3 MHz (7.3ppm deviation)
    Interpretation:
    - crystal frequency error is factory calibrated (my 7.3ppm are a device specific value)
    - DDS-generator is frequency corrected, frequency counter is not corrected

2) Frequency Resolution
    - expected 10mHz resolution at any frequency
    - use the setup for "Frequency Error" and try to set the frequency in a way that the
      counter is showing 20.000 000 0 MHz
    -> you will find that the frequency resolution is in the range of 1...5Hz (in an irregular fashion)
    Interpretation:
    - frequency correction is running into numerical issues
    - required number of discrete frequencies is 25MHz/10mHz = 2.5 x 10e9
      would nicely fit into 32bit unsigned integer
    - my gut feeling
      frequency correction is using 32bit floating point format (with 25bit mantissa)
      -> missing 7bit -> 10mHz x 128 = 1.3Hz  b.c. resolution @ max. frequency

3) deterministic square wave distortion
    - set generator to 1kHz square wave
    - connect a scope, trigger on falling edge
    - zoom in (e.g. 1us/div)
    -> you will observe a very surprising distortion!
        a small "bonus step" 1us in front of final falling edge
        height of step is approx. 12% of pp signal amplitude
        length of distortion is a function of frequency (approx. 0.001 * 1/f)
    Interpretation:
           I have no glue !!!

4) temperature of output driver ICs (at least for the replacement type)
    - they are really hot!
    -> strongly recomment to invest in a bigger heat sink and a small amount
         of thermally conductive paste


My general impression:

      good enough entry level product at an attractive price point
      (even with the known impairments)

Have Fun !

[Kleinstein]

With just R2R resistors at the FPGA output, is there a really improvement going from 8 to 12 bit - even the 8 Bit are difficult to get right. So how does a lower frequency triangle look like ?

The slew rate limit is not that bad for a sine wave. Reducing the amplitude at the highest frequencies is not such a bad restriction. It is more a problem to the square wave, but here using just an NCO there is a lot of jitter anyway. So not a good square wave anyway. A good square wave would need a filter and comparator.

Wondering why the use the extra AD8017 and still the AD812 for the output. They could have used the same OPs for both function. Likely the AD8017 would be slightly better for the output stage as well.

[nickcres13]

I posted this same reply on the other relevant MHS5200A teardown/hack thread, but thought I would post here for those not in the other thread...sorry for the spamming haha.

I know im late, but thought I would put in a word about some scripts im developing to further gain access to the MHS5200A device via serial commands. Namely, I have started with a dumb script that asks it for the frequency (on "EXT_IN") once a second and plots it in gnuplot. Links to the github repo and a boring video of me using it below. Let me know what needs fixed!

NOTE: I compile and run on Ubuntu, should work on other Linux distros, as for Windows and hackcoughblechApple there is work to do still obviously...

https://github.com/electricsheeplabs/mhs5200aDataLogger

[techfan]

Hi, I have a question, what is the maximum continuous current (square wave 100%) that supports the equipment ?? Can be used as DC source for Arduino or Raspberry Pi 3 ??
Thanks

[technogeeky]

Hi, I have a question, what is the maximum continuous current (square wave 100%) that supports the equipment ?? Can be used as DC source for Arduino or Raspberry Pi 3 ??
Thanks

No way. It's something like 20 to 30 mA. Even if that's off, it's still a few orders of magnitude too little current.

[techfan]

Ok thanks, I was not aware that it supplied so little current.

[karkoon]

Hi all,

New to the forum and first post here. Recently ordered MHS-5200A. I am not an expert of the electronics. Learning my way through the jungle of electronics here.

I have a noob question to this forum, can you generate an AM wave using MHS-5200A? If yes, can you explain how?

Thank you.

[nickcres13]

I think your options are arbitrary wave function output (so it would be a fixed AM wave you create) or mixing two outputs, one being carrier and the other is signal... so the answer is "no", but also yes.

[gico750]

guys I have had this tool for a few years and I would like to change the chips to improve the output signal, I show you some photos of the interior because in my model there is a dissipated voltage regulator complete with a cooling fan.
the external power supply is a 12v. 3A. that arrives in this internal regulator which then supplies the voltages to the device.
instead the two chips of the first and second channel have to be changed, besides I don't see any filter like in your photos, what do you think?