RIGOL MHO98 and MHO900 Oscilloscope Series

[Martin72]

Here are the adjustable bandwidths 20 MHz and 250 MHz.
The fact that it is almost exactly 250 MHz may be a coincidence, but the 20 MHz bandwidth is within the tolerance range that can be expected from a passive filter.

[kmoonwalker]

I was curios about the noise. Rigol says, in the datasheet, that the 12bit scope has an ENOB of 8 bits, so does that mean that they're cheating in the listings or that their 8-bit scopes are only 6 ENOB or something like that?

Also, the noise floor (50ohm) is 406.8uV at 10mV/div (the closest I could get in the datasheet to their recommended 8mV/div), which seems a bit high for 12-bit or even 8-bit unless they increased the number of divisions on the screen. Like, 0.08/(2^8) is ~312.5uV. Like, am I missing something here?

EDIT: I was really excited about this scope, but now that I read the specs, I'm rather tepid.

well it is always something for something and this is only one with such specs to price ratio which makes this even harder so some comprimises are to be expected (and not small ones most probably - we already seen remarks about input)

about scopes ADC capabilities and ENOB score it is perfectly normal for almost any equipment making such conversion to cut bits for sake of computing power/memory/pcb complexity, internal heating, etc savings

every bit chopped off means far less processing power necessary and scope need really lot of it

easiest proof is wfms/s parameter in real time observation - most of cheaper side to 5000series of Rigol platform is 30-50k only in real time while competition made it a bit faster (in best conditions of course) but still this is big difference why Siglents 3000x HD 1GHz model cost so much more, processes best case 200k wfms/s (so over 6 times more and equiv price) and has only like 0,2 more enob or something. Magnova on the other hand sacrificed how many Gsamples per second it could get to keep up 16 bit processing (thou I do not think that there was a teardown so far so we are basing on datasheet)

[egonotto]

Hello,

Does the MHO98 really have 250 MS and 2 GS/s for the digital channels? And do you also have that when you use the analog channels?

Best regards,
egonotto

[0xdeadbeef]

The 250Mpts assumption is based on the file size change of a binary file with/without the LA activated and assuming that each sample takes 16bit. The number of samples could be verified 100% e.g. by looking at the waveform headers.
Not sure where you got the 2GSa/s from. I would have assumed 1GSa/s. Then again, if the number of samples would be 100% sure, the period/time of the stored data would make it possible to determine the sampling rate.
Maybe also a CSV file could more give insight.

[sdouble]

For frenchies : any clue where I could buy in France, short term some MHO98 for business purposes ?

[eurofox]

For frenchies : any clue where I could buy in France, short term some MHO98 for business purposes ?

Amazon.fr 

[mironex]

I would like to ask about the MHO900 series.

1. Does it make sense to buy it based on software quality?
It seems comparable to the MSO5000 (for example, it appears to use the same logic probe).
In my opinion, the MHO900 series offers a good price for a 14-bit ADC and 4 GSa/s sampling rate.

2. Is it possible to hack the MHO900?

What is your opinion? Thank you! :-)

[smk]

The 250Mpts assumption is based on the file size change of a binary file with/without the LA activated and assuming that each sample takes 16bit. The number of samples could be verified 100% e.g. by looking at the waveform headers.
Not sure where you got the 2GSa/s from. I would have assumed 1GSa/s. Then again, if the number of samples would be 100% sure, the period/time of the stored data would make it possible to determine the sampling rate.
Maybe also a CSV file could more give insight.

A full 500MS file would be too big but the much smaller 1kS files should have the same structure.
The .zip file contains .bin and .csv files of a 10MHz -1V square wave on channel 1 and a 230MHZ square wave on LA D1 in the larger files.

[Martin72]

Hi mironex,

The MHO900 does not have a 14-bit ADC, but a 12-bit ADC.
I think it's quite comparable to an MSO5000, if only because of the two AWGs on board and the logic inputs.
The screen is smaller, but this is partly compensated for by the “full screen” function.
What I like at the moment is that Rigol has finally “boosted” the FFT function, even if there are still one or two minor issues (no manual selection of memory points, no markers).
But I'm still testing it, although I can already say that the DHO900 is “burnt” in comparison.
I said early on that the DHO900 series is a chimera; the MHO900 is the real upgrade to the DHO800.

[Noy]

Is MHO98 / MHO900 a better option to MSO5000 in comparsion to mostly digital stuff?
With >=2 Probes parallel used...

[Fungus]

I would like to ask about the MHO900 series.

1. Does it make sense to buy it based on software quality?

Yes.

2. Is it possible to hack the MHO900?

It's 99.9% probable but I haven't seen anybody try it yet. So far everybody's buying the high-end MHO98 model.

[Sorama]

I would like to ask about the MHO900 series.

1. Does it make sense to buy it based on software quality?

Yes.

2. Is it possible to hack the MHO900?

It's 99.9% probable but I haven't seen anybody try it yet. So far everybody's buying the high-end MHO98 model.

??
Software quality?
What quality? You mean the lack of it.

99,9% probable?
How do you know?
No, you don’t.

Everybody buying the limited 98 version?
How do you know?
Again, you don’t!

 

[Fungus]

For frenchies : any clue where I could buy in France, short term some MHO98 for business purposes ?

Amazon.fr 

I think he meant loaner/rental units. 

Buying these numbered, limited edition models on Amazon with the INTENTION of using/abusing them for business then returning them would be mean in the extreme.

There'll probably be rental MHO984s soon though, which is the same device for practical purposes.

[0xdeadbeef]

A full 500MS file would be too big but the much smaller 1kS files should have the same structure.
The .zip file contains .bin and .csv files of a 10MHz -1V square wave on channel 1 and a 230MHZ square wave on LA D1 in the larger files.

Looking at the csv files: tInc = 2.500000e-10 -> 4Gsa/s. Only every forth line, there is an LA value -> 1GSa/s.
A quick look into the bin files shows that each 16bit data point actually needs four bytes.
Even more surprisingly, while the Waveform Header starting at offset 0x104c says that there are 250 logic values stored (which matches the CSV), the Waveform Data Header starting at offset 0x10d8 says that the number of bytes in the buffer is 4000, which would mean 1000 entries of 4 bytes. But there are actually only 1000 bytes stored. Which again fits the 250 values.

-> looks like the LA samples with 1GSa/s and the sample memory is 125MPts since each 16bit sample in the binary files is stored in four bytes (for whatever reason).

[Fungus]

<snip>

Please do yourself a favor and don't buy one of these, Sorama.

Leave them to everybody else.

[egonotto]

Hello,

thank you for the information.
So the analog data is sampled at 4 GHz, but the digital data only at 1 GHz. That means there are only 250 samples for the digital data.
I assume that the digital channels have a maximum sampling rate of 1 GHz and 125 MSample memory.
I have no idea what else is stored in the binary.

Best regards,
egonotto

[0xdeadbeef]

I assume that the digital channels have a maximum sampling rate of 1 GHz and 125 MSample memory.
I have no idea what else is stored in the binary.

Yes, the sampling rate for the LA is 1GHz and the sampling memory is 125Mpts. But - for whatever reason - each 16bit sample is stored as 4 bytes in the binary format.

[Sorama]

Leave them to everybody else.

If you don’t have one already, you’re too late for the party.

[0xdeadbeef]

If you don’t have one already, you’re too late for the party.

I don't think so. I would think they haven't even reached 1000 sold units yet and there is a new shipment on its way to Europe which is supposed to arrive in two weeks or so.
Buying directly from Rigol (through their website or Amazon) should still work.

[Martin72]

Hi,


Fan noise....In "real" it´s more annoying.

[Sorama]

If you don’t have one already, you’re too late for the party.

I don't think so. I would think they haven't even reached 1000 sold units yet and there is a new shipment on its way to Europe which is supposed to arrive in two weeks or so.
Buying directly from Rigol (through their website or Amazon) should still work.

There are probably lots of backorders.
Not sure if you can still order one.
At least not from the well known distributors.

[0xdeadbeef]

Well, at least through Amazon, they still take orders.
On their own site, they don't take order currently but stated on the 16th of October that a new shipment is on its way to Europe.

[Sorama]

Well, at least through Amazon, they still take orders.
On their own site, they don't take order currently but stated on the 16th of October that a new shipment is on its way to Europe.

And that next shipment is probably to fulfill backorders.

[Martin72]

Hi,
After the update rate test, I played around a bit more...
The hidden test mode is also available here, using the familiar method to access it.
I took a few screenshots of it.

[Martin72]

Finally, update rate test...

Setup:



I connected the Magnova to the trigger output of the Rigol, which then measured N-Counts.
The time base was 100 ms, so the counter values directly refer to 1 second.
In addition, there was the frequency measurement.
Rigol specifies a rate of 30000 wfms/s for the MHO900 in normal mode. N-Count is quite accurate, while the frequency sometimes appears breathtakingly high.
Therefore, in this case, I place more trust in the N-Count function.
The constancy in the first time ranges is remarkable before the rate drops as expected.