R&S RTM3004 The Swiss-Knife-Oscilloscope

[mikeselectricstuff]

My biggest complaint about the R&S scopes is the slightly sluggish UI, when compared the Keysight. I was a bit disappointed the 3004 didn't improve on this.
Compared to others it's probably fine, I've just been so used to Agilent then KS over the years.
I think there is scope to improve this by avoiding some of the transitions etc. when menus appear and disappear, but it does seem to miss button-presses sometimes.
I'll probably use the 3004 as my day-to-day scope for a while to see how it feels in everyday use.

Oh, and those damn clicky encoders...
   

[nctnico]

Problem is, despite the shortcomings (memory, noise, can't apply decodes after an acquisition) keysight has a scope with all the options others have only in higher class scopes.

What kind of options?

[JPortici]

The whole set of decoders!
I hope that the fact that they released this scope with more than the usual five (SPI,I2C,UART,CAN,LIN) is a sign that they will add more in the future, but who know what will really happen?
to me this is more important than having a gazillion sample memory depth

[mikeselectricstuff]

The whole set of decoders!
I hope that the fact that they released this scope with more than the usual five (SPI,I2C,UART,CAN,LIN) is a sign that they will add more in the future, but who know what will really happen?
to me this is more important than having a gazillion sample memory depth

What other decoders would you want? USB and Ethernet are the only ones that come to mind.
bear in mind you can pull waveform data off quickly and easily via USB, so offline analysis should be doable.
Of course what would be really nice is an architecture where you could write your own.

[GeorgeOfTheJungle]

I'm looking at it here: https://www.batronix.com/shop/oscilloscopes/Rohde-Schwarz-RTM3004.html

RTM3004 -> 4510€ (minimum with 4 channels)
RTM-PK1 Option Bundle -> 2975€ (triggers, decoders, etc)

TOTAL: 7485€

Is that right? Or is there any launch campaign or bundle or something somewhere?

[nctnico]

The whole set of decoders!
I hope that the fact that they released this scope with more than the usual five (SPI,I2C,UART,CAN,LIN) is a sign that they will add more in the future, but who know what will really happen?
to me this is more important than having a gazillion sample memory depth

The Keysight 3000T seems to support these decoders according to the datasheet:
I2C
SPI (2/3/4 wire)
RS232/422/485/UART
CAN / CAN-dbc /CAN-FD (CAN-FD ISO)
LIN / LIN symbolic
SENT
CXPI
FlexRay
MIL-STD 1553
ARINC 429
USB PD
I2S
User-definable Manchester / NRZ

That is quite an impressive list. Ofcourse it depends on what you are working on whether you would need these decoders or not. To me user definable Manchester / NRZ would be handy every now and then.

[mikeselectricstuff]

I'm looking at it here: https://www.batronix.com/shop/oscilloscopes/Rohde-Schwarz-RTM3004.html

RTM3004 -> 4510€ (minimum with 4 channels)
RTM-PK1 Option Bundle -> 2975€ (triggers, decoders, etc)

TOTAL: 7485€

Is that right? Or is there any launch campaign or bundle or something somewhere?

For 350MHz and up there is a bundle with a free spectrum analyzer :
https://www.testforce.com/news/promos/free-spectrum-analyzer.html

[mikeselectricstuff]

BTW I was surprised that there is still no ability to trigger internally from the wavegen/wavegen sweep

[GeorgeOfTheJungle]

USB PD

That's USB 1.0 (USB low-speed (1.5 Mbit/s) and full-speed (12 Mbit/s)), yes? How much does a 3000T with all these triggers/decoders cost? What kills me is the 4M max mem depth, that's less than the (older) infiniivision I have now.

[hwj-d]

BTW I was surprised that there is still no ability to trigger internally from the wavegen/wavegen sweep

Is trigout-aux stable/reliable now?

[3db]

Nice review Mike.
Thanks for the effort.
3db.

[mikeselectricstuff]

BTW I was surprised that there is still no ability to trigger internally from the wavegen/wavegen sweep

Is trigout-aux stable/reliable now?

It still seems rather erratic , with variable-sized gaps in the bursts, and the time between triggers being rather longer than expected - not looked in detail yet.

[nctnico]

BTW I was surprised that there is still no ability to trigger internally from the wavegen/wavegen sweep

Is trigout-aux stable/reliable now?

It still seems rather erratic , with variable-sized gaps in the bursts, and the time between triggers being rather longer than expected - not looked in detail yet.

AFAIK it is not uncommon for DSOs to have a variable trigger rate. At some point the data will need to be processed and depending on mode/settings/math the interruptions will be longer or shorter.

[TheSteve]

Thanks for the tear down Mike - the scope looks to be very nicely built. The front end amplifiers needing such a heatsink surprised me though. The equivalent part on the Keysight 2000/3000 scopes has nothing.
Could this be a product of them trying to go for lower noise because of the 10 bit ADC?

[mikeselectricstuff]

Thanks for the tear down Mike - the scope looks to be very nicely built. The front end amplifiers needing such a heatsink surprised me though. The equivalent part on the Keysight 2000/3000 scopes has nothing.
Could this be a product of them trying to go for lower noise because of the 10 bit ADC?

Hard to say.
May also be for stability, keeping temps as low as possible. With 500uV/div, DC drift is probably a significant issue.
I measured the temp on the big FPGA sink and it was only about 40 deg.C so the HS was overkill but keeps temps down and means they can use a quiet fan.
 

[hwj-d]

BTW I was surprised that there is still no ability to trigger internally from the wavegen/wavegen sweep

Is trigout-aux stable/reliable now?

It still seems rather erratic , with variable-sized gaps in the bursts, and the time between triggers being rather longer than expected - not looked in detail yet.

AFAIK it is not uncommon for DSOs to have a variable trigger rate. At some point the data will need to be processed and depending on mode/settings/math the interruptions will be longer or shorter.

Thanks Mike for your nice competent work. 

Thanks ntcniko for your answer.
Maybe I just expect too much from such a modern scope. Perhaps in practice, it is more likely to be able to receive any trigger event for, for example, synchronization purposes, than precisely each one. So, it is more like a slide rule, than a precise counting calculator (edit: the trig-aux out, not the scope), right? If it's so, than that's an interesting insight for me. 

[mikeselectricstuff]

I assume that the trig-out really does output a pulse exactly on every trigger event, but it would be interesting to check that this is actually the case. I have some ideas of how to test this definitively.
Something that definitely does happen is that on each trigger it captures data both to the left and right of the displayed screen, so the shortest trigger-to-trigger time is always significantly more than one screenful. There may be situations like catching occasional glitches where it would be useful to change this behaviour to only do one screen-time's worth of data before re-arming.

[Someone]

I assume that the trig-out really does output a pulse exactly on every trigger event, but it would be interesting to check that this is actually the case. I have some ideas of how to test this definitively.

The confusion arrives because the trigger output on most scopes is a replication of the acquisition trigger, which is why the trigger rate jumps up when switching to a (ideally circular) segmented acquisition mode as its not (generally) processing the acquisitions and has shorter dead time.

[hwj-d]

I assume that the trig-out really does output a pulse exactly on every trigger event, but it would be interesting to check that this is actually the case. I have some ideas of how to test this definitively.
Something that definitely does happen is that on each trigger it captures data both to the left and right of the displayed screen, so the shortest trigger-to-trigger time is always significantly more than one screenful. There may be situations like catching occasional glitches where it would be useful to change this behaviour to only do one screen-time's worth of data before re-arming.

Ok, got it. Thanks again.

The confusion arrives because the trigger output on most scopes is a replication of the acquisition trigger, which is why the trigger rate jumps up when switching to a (ideally circular) segmented acquisition mode as its not (generally) processing the acquisitions and has shorter dead time.

Yes, that's also reasonable. Thanks too.

[maginnovision]

I tested my RTB2004 and the trig out appears to be acquisition-based, but if you use the counter and set it to trigger it is actual triggers. I assume the 3K will be the same? I just checked with a 24MHZ clock. Used 1252B freq counter to read trig out from AUX out port, and verified by changing memory depth. Bigger depth, lower AUX trig out frequency but more stable Trigger Counter output, lower depth the opposite.

[ws2812b]

I summarised the differences from the datasheets earlier in this thread, and have subsequently edited for the things that were wrong in the datasheets.
First vid should be done sometime over the weekend.

thanks for the vid. Things like pulling out the menu was exactly what I meant,
@Rich: do we get this for the RTB2k as well?

[MikeP]

 Rich.
 I see improved noise performance. Are the distortions of the input path reduced in comparison with the RTB?

 Mike, great video. Thank you.

[TheUnnamedNewbie]

Thanks for the tear down Mike - the scope looks to be very nicely built. The front end amplifiers needing such a heatsink surprised me though. The equivalent part on the Keysight 2000/3000 scopes has nothing.
Could this be a product of them trying to go for lower noise because of the 10 bit ADC?

Could also be they are using a different process technology with a bit more noise and need more current to offset it? Since power isn't critical in this application it could be easier for them to just up the current (and thus power) instead of having to jump through hoops to get the noise down without increasing power.
Also seems like the noise performance is better than the competition (it has to be to get the most out of that 10 bit ADC on the low ranges).

[Rich@RohdeScopesUSA]

I summarised the differences from the datasheets earlier in this thread, and have subsequently edited for the things that were wrong in the datasheets.
First vid should be done sometime over the weekend.

thanks for the vid. Things like pulling out the menu was exactly what I meant,
@Rich: do we get this for the RTB2k as well?

Yes, it is planned in a future release.

-Rich

[Rich@RohdeScopesUSA]

Rich.
 I see improved noise performance. Are the distortions of the input path reduced in comparison with the RTB?

 Mike, great video. Thank you.

The 3000/4000 use a new front-end that is different from the 2000's.  While the 2000 has excellent noise for the its class, the 3000/4000 are even better while still offering full bandwidth (up to 1GHz) at all volt/div settings, even 500uV/div. 

-Rich