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

[SaabFAN]

I just did some reading on the subject: Apparently, only inverting configurations can be used as mixers to properly sum voltages, which, I guess, is necessary in this case to add/subtract an offset from the DAC-Voltage. It is then fed into the variable gain Amplifier, and then again inverted to give the correct output-polarity. Actually not a bad idea, but in this case not up to the task.

What (I think) they should have done: Mix the signals with the first inverting amp, invert the signal again (to have correct polarity), feed it into the variable gain amplifier and then drive an amplifier configured as a non-inverting amp.
Would have added two ICs (or one if one doesn't care about crosstalk too much^^) to the BOM, plus some passives, and required some more space on the PCB.
I guess, this is how they do it on the High-Power Option that is available for these units, which fits in the large empty space next to the PCB inside the case.

[devttys0]

That's true, but I didn't see any summing amplifiers in the circuit. From what I've seen, they achieve the DC offset by adjusting Vref on the non-inverting input of the first AD812 op-amp; I don't see any reason why they wouldn't be able to do that in a non-inverting configuration, but perhaps someone can enlighten me.

[SaabFAN]

Refitting a Battery and a boost-converter didn't help much, though it appears as if the square-wave at 1 and 2Mhz looks a little bit better...
Installing the Battery was more difficult than anticipated btw. The signal-gen needs about 6A for almost 11ms, before the current-consumption goes down to about 250mA. It can increase up to 800mA with both channels terminated into 50Ohms at 15V and 12Mhz. I therefore had to modify my original circuit quite a bit - First putting in large caps and after that didn't work, I added coils. Now the unit powers on with a 80% charged battery or if the external PSU is connected. I probably could improve that by adding a SuperCap, but I don't have one here and don't plan on buying one specifically for this unit.
I estimate the cost to modify the device in this manner to be about 15€, including shipping.

Attached: Screenshots of my measurements.
All Waveform-Measurements done with 15V Amplitude
Sine-Waves:
Quickprint22 - +15% Offset, 75Ohms
Quickprint23 - 0% Offset, 75Ohms
Quickprint24 - 0% Offset, 1MOhms
Square-Waves: All measurements with 0% Offset
Large Amplitude - 1MOhms
Small Amplitude - 75Ohms

[SaabFAN]

And here are some pictures of the modified unit + Current-Draw when turning on (Measured with Scope-Probe over the Current-Shunt of my Multimeter, PSU: KA3005P)

Quickprint18: Current-Draw
Quickprint19: Voltage-Drop

The last 2 Pictures show the first, rather clean version of the Battery-Conversion and the quite messy one with all the modifications necessary to make it work^^

[devttys0]

I managed to significantly improve the sine wave distortion and square wave overshoot/ringing with a couple of simple fixes:



The first was to replace the AD812 with a THS3092. They're pin compatible, but the THS3092 has a much higher slew rate, and I can now output a 15dBm sine wave at 25MHz and the harmonic content is actually within spec (highest harmonic is at -33dBc).

If you look at the attached HarmonicDistortion image, the output of the AD812 is shown in the purple trace, while the output of the THS3092 at the same frequency/power is shown in yellow. The yellow trace is a bit hard to see since it's underneath the purple trace, but you can just make it out; significant improvement!

I confirmed that the ringing/overshoot was caused by the elliptic low pass filter used on the output of the DAC. I replaced it with a Bessel filter designed for a 3dB cutoff at 30MHz, and there's practically no ringing on the waveform. The SquareWave1 and SquareWave2 images show the Bessel filter output on channel 1 (yellow) and the unmodified filter output on channel 2 (green).

The noise level on the output of the Bessel filter is a bit higher when viewed on the spectrum analyzer, especially near the carrier, but it's not too much worse and to be honest the noise performance of this function generator didn't seem to be all that great to begin with. I decided to stick with the Bessel filter for the cleaner square waves, but a Butterworth might split the difference in performance between the Bessel and elliptic filters.

[SaabFAN]

Nice one, I'm ordering parts right now to try the mod myself
I'm curious why they used the AD812 instead of the THS3092 if it is more expensive. Maybe you should send them a link to your videos and show them how it's done (They may even give you their other Signal Gen for free to do some improvements there too ).

About my portability-mod: It is unfortunately necessary to use a 20C-Rated Battery to deliver enough PowerUp-Current. Something, I don't know what, loads down the rails and it is not capacitors, because the rails stay pulled below 2.5V even after a minute or so.
The overall price of modifying the unit to be portable is about 15 to 20€, depending on the battery-size. A little bit more if a 7.4V Battery is used because a different charger and a different Wall-Adaptor are required.
A 1000mAh lithium Battery is able to power the signal-gen for about 30 minutes with both channels terminated into 50 Ohms and outputting 1MHz Square-Waves @ 15V. Quite handy to lose yet another cable (Bench-Multimeter is also on Batteries) on my already crowded bench which actually is my Desk that is also occupied by my PC-Monitor and a 4-channel Audio Mixer

Would you say that with your modifications, this Signal-Gen has actual value and could be considered a bargain, or is it still highly preferable to spend 200€ more to buy a unit from Rigol/Siglent/Owon with similar/inferior specs in terms of maximum sine-frequency?

[devttys0]

I'm curious why they used the AD812 instead of the THS3092 if it is more expensive

All else being equal (unless I'm missing something), the THS3092 does seem like the better choice here. 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.

Would you say that with your modifications, this Signal-Gen has actual value and could be considered a bargain, or is it still highly preferable to spend 200€ more to buy a unit from Rigol/Siglent/Owon with similar/inferior specs in terms of maximum sine-frequency?

With or without the mods, I'd say it's still a pretty good deal for the price, just as long as you're aware of its limitations. With the mods its much more acceptable, at least for my needs, but it still has other issues (e.g., square wave jitter is still pretty bad, especially at certain frequencies).

If you want a good, trustworthy function generator, spend the money on a good one (I got an older HP/Agilent off eBay, it's great). The MHS-5200 is useful as a secondary function generator when I need more than one signal, or if I need two phase-locked signals, etc.

[motocoder]

I know it's a more extensive mod, and maybe not do-able without the source code to the micro, but it seems like the output filter stage should be totally removed for square wave output.

[krivx]

I'm curious why they used the AD812 instead of the THS3092 if it is more expensive

All else being equal (unless I'm missing something), the THS3092 does seem like the better choice here. 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.

This seems likely if they are also using it in another product.

[usagi]

I'm curious why they used the AD812 instead of the THS3092 if it is more expensive

All else being equal (unless I'm missing something), the THS3092 does seem like the better choice here. 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.

or it could be a counterfeit ad812 which they get for a tiny fraction of the price. sorta like all the counterfeit lm2596 out there.

[wd5gnr]

Nice one, I'm ordering parts right now to try the mod myself

Would you (or OP) consider posting a BOM?

I assume the THS3092 is the one without the bottom pad, for example. And the SMD inductors and caps... from the video:
6.5 30 51 and 130pf -- what are those maybe 0402?
And the L are 160, 330, and 560 nH in the same case?

Then 2X everything.

Great job and great video!

[krivx]

Nice one, I'm ordering parts right now to try the mod myself

Would you (or OP) consider posting a BOM?

I assume the THS3092 is the one without the bottom pad, for example. And the SMD inductors and caps... from the video:
6.5 30 51 and 130pf -- what are those maybe 0402?
And the L are 160, 330, and 560 nH in the same case?

Then 2X everything.

Great job and great video!

Looks like a BOM has been posted here http://www.analogzoo.com/2015/08/fixing-the-mhs-5200a/

[devttys0]

Yup, as krivx pointed out, there's a digikey BOM in my blog post (CSV format, should just be able to upload it to digikey). There are also some more detailed pictures/descriptions of the mods there too.

The caps/inductors I ordered were mistakenly 0402's, while the pads on the board are 0603, so you might want to change them if you aren't comfortable working with 0402 (either size fits on the pads fine).

[wd5gnr]

Thanks! Yeah I got the gear to do worse, but my eyes probably like the 0603s better ;-)

[TheRadioGeek]

I have just finished the mods posted by Devttys0  [http://www.analogzoo.com/2015/08/fixing-the-mhs-5200a]and I just wanted to thank him for this nice improvement. I did order the 0603 size parts because they are easier to work with and fit on the pads better. I have a few before and after mod screen shots below.

[Wob_Bidlar]

Hey!
Just a quick update on the MHS-5200-A:

A month ago I bought one of these; today I received it and wanted to do these cool mods (thx devttys0 for your great work) but when I opened it I found out that they (the manufacturer) "modified" the Low-Pass filter for the DAC!! 
This is what it looks like (I didn't change anything!!):

[gr8tfly]

Hey!
Just a quick update on the MHS-5200-A:

A month ago I bought one of these; today I received it and wanted to do these cool mods (thx devttys0 for your great work) but when I opened it I found out that they (the manufacturer) "modified" the Low-Pass filter for the DAC!! 
This is what it looks like (I didn't change anything!!):

I received mine about a week ago (25 MHz version) and the output looked pretty nice. So, knowing about the design tweaks and seeing your post, I cracked mine open to have a peek. Lo and behold, mine had been modded at the factory in almost an identical fashion. (There are three solder blobs across the components instead the long line yours has – see attached).

On a side note: for fun, I fed the output back into the counter input. Perhaps unsurprisingly, they don't agree. At 11 MHz, my freshly calibrated receiver (< .1 ppm to 10 MHz WWV on an ICOM 756 Pro-III), showed 5200A output was about 50 Hz off, which isn't too bad. The counter was off a bit more, and in the opposite direction. Though obviously not built as a lab quality instrument, it would still be fun to at least "trim" it and see how it tracks over time. Just something else to fiddle with, which is what all this stuff is for, right? 

My question is whether anyone has any info on the "internal calibration" mentioned in their manual. I'm hoping there would be another set of commands for calibration, or at least some front panel key press voodoo to invoke it.

[danadak]

I loaded this new version (link below), connected gen, and tried to do an arbitrary waveform.
Loaded it on "Arbitrary" tab, then wrote to arb00 mem location.


https://www.dropbox.com/s/cdlsuv2w5rkss6g/MHS-5200A_en5.0.rar?dl=0


It shows up on "Control Panel" tab when I select arb00 as wavefom,
but does not output out of the generator.

I need a drink........

Also sweep mode do. And no sweep trigger out of TTL outputs on gen to synch the display
on scope to the sweep. Although I think one could use 2'ond channel to create the trigger
as both channels operate off same time base.

Regards, Dana.
Modify message
« Last Edit: May 01, 2016, 04:19:09 AM by danadak »

[pascal_sweden]

Why not go for the FeelTech FY3224S?

The DA convertor has a better resolution!

https://www.eevblog.com/forum/testgear/feeltech-fy3224s-24mhz-2-channel-dds-aw-function-signal-generator/

[danadak]

I purchased the Ming He last fall.

Still need to get this software to work -


https://www.dropbox.com/s/cdlsuv2w5rkss6g/MHS-5200A_en5.0.rar?dl=0


Regards, Dana.

[NJO]

My MHS-5200A 25MHz arrived yesterday. When I had a look inside I noticed a modified layout. Somebody knows if the described mods are still necessary or not? I have attached two pictures.

Regards,
Norbert

[NJO]

Installing the software (rar-file from here) I noticed that its not the MHS-5200a but the MHS-5225a function generator.
Is there any mod for it or are the mods already included there?

Regards,
Norbert

[Dwaine]

I just got my MHS-5200A in the mail and it has the redesigned PCB.   The THS3092 part number from mouser is it 595-THS3092D?  Also, could you use the 595-THS3095D?  The slew rate is higher for the 595-THS3095D

[Dwaine]

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

[pascal_sweden]

That's why you should go for the FeelTech FY3224S instead

https://www.eevblog.com/forum/testgear/feeltech-fy3224s-24mhz-2-channel-dds-aw-function-signal-generator/