With a budget ceiling of $7500.00 I don't think this is a technical discussion.
You would be better served by a cost / needs analysis focusing on your expected needs defined by a performance envelope for your DUT's.
For example, if your MicroProcessors are 16 Mhz, (I use 180 Mhz Teensy, so I would use 180 Mhz for example)
- Extreme case probing clk line 16 Mhz
- Then the very worst case ceiling on pins is is 8 Mhz on a pin, 1 clk HIGH, 1 clk LOW Generally overheads make this unreachable.
- List the rest of your circuit elements here.
- Capture the worst cases and select the highest one and you have your basic bandwidth. Determine the Nyquist / oversample factor you want. Usually 5X to 10X, some folks use 3X it depends on your sensitivity to accuracy / aliasing.
Envelope = basic bandwidth X oversample Factor.
Final Sample rate Envelope = Envelope X future proof factor ( I use 2, to account for a 360 Mhz Teensy within life of scope), but you pick the number you want)
How often will you hit your envelope / sample rate? % of time, 75%, 50% of envelope etc ( design the table to meet your needs )
How important is it to you (or your product) that you execute tests at the edges of the envelope? How much are you prepared to pay for the rare cases at the edge?
(Example: My worst case final envelope would be 180 X 5 X 2 = 1.8 Gs/sec)
Observations
1) Anything above 360 MHz is wasted.
2) Anything above 1.8 Gs/ sec is not worth paying for, but might be nice to have (note this implies a 70-100 Mhz scope).
3) In my case, the highest expected "need" would be a decode on a 100 Mhz SPI bus. But, this would be very rare << 1%
I constructed a table and found the following worst cases:
- Bit Bang (180 Mhz) - 0.9 Gs/sec per channel (highly unlikely to implement)
- SPI (100 Mhz) - 0.5 Gs/sec across 4 channels = 2 Gs/s total
- I2C (5 Mb/s) - 0.025 Gs/sec across 2 channels = 0.5 Gs/s total
- UART (.4608 Mb/s) - Not significant
- Can Bus (1 Mb/s) - Not significant
Next features:
Decodes = Mandatory
Number of channels = 4 - I want 4 so I can decode 4 wire SPI and 4 wire proprietary busses
AWG = Yes , I just want it. No real justification except conserve desk space which is a premium for me
Touch Screen = nice to have
MSO = Yes, most of my work is digital
Memory = This ended up being a HUGE factor, more below
I/O (LAN USB etc) = Nice to have
Big Screen = Important, I am getting old.. eyes getting worse.
Trigger Functions = Important but mid priority
Math Functions = Not sure, but don't want to ignore so yes
Masking = Not important
Ability to do Bode Plots = nice to have
FFT = Low priority
Other.
So my dream scope would have ~350 Mhz and a sample rate of 2 Gs or better and the features above
BUT a review of scope pricing placed a market value of > $4000 (US) and closer to $7k for a full featured scope, most of these came with large & segmented memories. BUT my budget was not $7.5k, it was closer to $1.2-1.5 k so I had to park the dream and cost optimize.
I looked at how often I would need things and what was most important to me, serial decode for signals in the 25 Mhz band became the most significant use case. (Note: I took a lot of effort and research to get to this point, but oddly the framework was similar to your stated needs)
At this price point 1.2k -1.5k, memory architecture
blows up as a very significant issue in terms of the capability / approach for long decodes.
I won't get into the details here but to meet my
specific needs the best compromise was either a used Keysight 3000 (I could not find one anywhere near my price point) or a 100 MHz Picoscope 3405D. Basically the memory architecture of all the other 4 channel MSO scopes in my price range prevented me from doing what I wanted to do.
AND Then I got lucky, I am almost never lucky! The introduction deal came for the R&S RTB2004 .. a little over my price point by $600 BUT it is very close to the "dream" spec. The rest is history.