Picoscope Hack

[_Wim_]

This is the schematic front end until first amplifier

EDIT: added resistor at offset input, as markone was correct, this caused confusion

[2N3055]

My 3406D was 2100€ few years ago when I bought it and worth every penny...
My 12000€+ Keysight MSOX3104T has 4MPts sample memory.

3406D MSO has 500Mpts...
I bought it for decoding and for high end features in software, SDK and FRA...
I also have 16bit Pico 4262 that is absolutely unique. At 1000€ it has specs no other scope has at any price..
And I also have 4824A that is 20MHz 12 bit 8 channel scope.. It is not 200€, but go find 8ch 12bit scope that is not 10x more. etc etc...

Fact is, my two 12 bit Siglents mostly replaced MSOX3104T in use, but not Picoscopes...
When it's time for Picoscope, it's time for it.

I agree that their pricing is not following the same logic as some other manufacturers. But it is special product in many ways..

[markone]

This is the schematic front end until first amplifier

I guess there is a resistor in between offset setting and opamp negative input.

10pF and 7pF capacitors were actually measured ?

[_Wim_]

I guess there is a resistor in between offset setting and opamp negative input.

Yes, I did not further draw that part => Edit => now updated in the schematic above

10pF and 7pF capacitors were actually measured ?

They were measured in circuit with my HP 4276A, so that will not be very accurate. For most caps I could not even get a reading, but for the onces I did, take them certainly with a grain of salt (I did not fance desoldering them all risking that one "jumps" away or is damaged in the process...)

[markone]

My 3406D was 2100€ few years ago when I bought it and worth every penny...
My 12000€+ Keysight MSOX3104T has 4MPts sample memory.

3406D MSO has 500Mpts...
I bought it for decoding and for high end features in software, SDK and FRA...
I also have 16bit Pico 4262 that is absolutely unique. At 1000€ it has specs no other scope has at any price..
And I also have 4824A that is 20MHz 12 bit 8 channel scope.. It is not 200€, but go find 8ch 12bit scope that is not 10x more. etc etc...

Fact is, my two 12 bit Siglents mostly replaced MSOX3104T in use, but not Picoscopes...
When it's time for Picoscope, it's time for it.

All the models that you listed are quite peculiar/unique, this does not apply to ordinary models, especially 2000 series, that i would consider overpriced for USB2.0 devices.

I was tempted for a 3000 series, but sweet spot SKUs are not available at all.

I agree that their pricing is not following the same logic as some other manufacturers. But it is special product in many ways..

That's my point, they are increasing prices also for old models when "some other manufacturers" are improving day by day specs and listing cost and often make sales campaign.

[markone]

I guess there is a resistor in between offset setting and opamp negative input.

Yes, I did not further draw that part => Edit => now updated in the schematic above

10pF and 7pF capacitors were actually measured ?

They were measured in circuit with my HP 4276A, so that will not be very accurate. For most caps I could not even get a reading, but for the onces I did, take them certainly with a grain of salt (I did not fance desoldering them all risking that one "jumps" away or is damaged in the process...)

One method would be desolder resistors placed in series, much less fragile and marked and easy to replace, but it's not worth doing it if you are not interested to perform an AC simulation.

[egonotto]

Hello,

I paid about 950 € for the PS 5243A in 2015 and about 1972 € for the PS 5444B in 2017.

Best regards
egonotto

[markone]

Hello,

I paid about 950 € for the PS 5243A in 2015 and about 1972 € for the PS 5444B in 2017.

Best regards
egonotto

5444B seems no more available, there is 5444D at 3700 euros ... (or even more, depending on the seller) 

[jasonRF]

All the models that you listed are quite peculiar/unique, this does not apply to ordinary models, especially 2000 series, that i would consider overpriced for USB2.0 devices.

Sure, they are expensive compared to budget benchtop scopes made by Rigol/Siglent/Instek/etc.  But early this year I was in the market for a USB scope upgrade where I wanted 2 channels and at least 100 MHz bandwidth, along with deep memory, large FFTs, serial decoding and either high-res mode or digital low-pass filtering.  I did not find any viable options besides Picoscope.  Sure, you can find cheaper USB scopes with reasonable hardware specs manufactured by Owon or DreamSourceLab or others, but you just do not get the rest of the capabilities.  My budget was too low for a 2208B (or even a 2207B), so I shopped ebay for a handful of months until I found something I could afford.  I would have been happy with a 2000 or 3000-series, but the first good option I found was a 5244B that I picked up for $450 US.

I think a lot of the Picoscope cost is due to design and manufacturing in the UK (at least both of mine are made in the UK), relatively low sales volumes, and the superior software/firmware.  If a company like Owon decided to be serious about their software, they could give Pico enough competition to force prices down, but this doesn't seem likely to happen any time soon.

jason

[Scratch.HTF]

Has anyone tried to upgrade the waveform memory (DDR2 SO-DIMM) in a PicoScope 6000 series (mine is a 6403B with 512 MB common memory; the 500 MHz model can have up to 2 GB common memory) and if so, does it require a code and/or different control software to enable the extra memory?
If the memory can be upgraded, is it also possible to upgrade beyond 2 GB using a special code or different control software if necessary?
From what I know, the PicoScope (with removable memory) uses SO-DIMM memory modules and there are no registered SO-DIMM modules out there (registered modules are typically required in workstations and servers with more than four DIMM slots and therefore, there was no need for registered SO-DIMM modules) - the pictures at https://www.eevblog.com/forum/testgear/picoscope-pico6402a-6404d-teardown/ show a Texas Instruments IC on the SO-DIMM which is most likely used for signal termination to prevent ringing at very high speed.

[markone]

All the models that you listed are quite peculiar/unique, this does not apply to ordinary models, especially 2000 series, that i would consider overpriced for USB2.0 devices.

Sure, they are expensive compared to budget benchtop scopes made by Rigol/Siglent/Instek/etc.  But early this year I was in the market for a USB scope upgrade where I wanted 2 channels and at least 100 MHz bandwidth, along with deep memory, large FFTs, serial decoding and either high-res mode or digital low-pass filtering.  I did not find any viable options besides Picoscope.  Sure, you can find cheaper USB scopes with reasonable hardware specs manufactured by Owon or DreamSourceLab or others, but you just do not get the rest of the capabilities.  My budget was too low for a 2208B (or even a 2207B), so I shopped ebay for a handful of months until I found something I could afford.  I would have been happy with a 2000 or 3000-series, but the first good option I found was a 5244B that I picked up for $450 US.

I think a lot of the Picoscope cost is due to design and manufacturing in the UK (at least both of mine are made in the UK), relatively low sales volumes, and the superior software/firmware.  If a company like Owon decided to be serious about their software, they could give Pico enough competition to force prices down, but this doesn't seem likely to happen any time soon.

jason

5244B at 450 USD seems quite a bargain, I guess that you would never have bought it at full price.

I agree to pay a plus for a professional SW support but to be honest Picoscope 6 nowadays sounds quite outdated, at that price mechanisms like serial protocol trigger and performance streaming mode should be a standard feature, on the other hand brands like Owon & Hantek, that sometimes provide decent HW for the money, are much worse, often at a level that renders their products almost useless for serious job. I do not mention Picoscope 7 because it's still almost work in progress.

Years ago, with lower prices and less competition from desktop DSO's market, Picoscope listing was much more justified, now you have to have quite peculiar/niche needing to find an acceptable balance money / features.

The fact that Picoscopes are assembled by third parts in UK it's a factor that increase cost but not at the that level, at least for that level of quantity.

You are right, if only Owon ... but i fear that sadly we never see that.

[_Wim_]

Woohoo, the bandwidth has been fixed! Opening the front end shield, there was a 75ohm resistor in series with the input. I have change this to a zero ohm resistor, and 

To further evaluate the frequency response at all ranges, I made the following test setup:

BG7TBL Noise source => DC-block => Stepped attenuator => 50-ohm input adaptor => Picoscope input

This gave me a relative smooth noise spectrum that I could attenuate, to test the response on all ranges. To make the comparison easier, all plots where normalized to the +-20V range (with zero ohm resistor mod). Goal was to compare the frequency response with different attenuator settings in the scope to ensure the zero ohm mod worked correctly with different attenuator settings.

The same test was performed with the unmodified input (with 75 ohm series resistor). Conclusion: both unmodified and modified had +-2dB variation over all the ranges.

[markone]

Woohoo, the bandwidth has been fixed! Opening the front end shield, there was a 75ohm resistor in series with the input. I have change this to a zero ohm resistor, and 

To further evaluate the frequency response at all ranges, I made the following test setup:

BG7TBL Noise source => DC-block => Stepped attenuator => 50-ohm input adaptor => Picoscope input

This gave me a relative smooth noise spectrum that I could attenuate, to test the response on all ranges. To make the comparison easier, all plots where normalized to the +-20V range (with zero ohm resistor mod). Goal was to compare the frequency response with different attenuator settings in the scope to ensure the zero ohm mod worked correctly with different attenuator settings.

The same test was performed with the unmodified input (with 75 ohm series resistor). Conclusion: both unmodified and modified had +-2dB variation over all the ranges.

Interesting, I would try with some other values for input resistor because with zero It would seem that aliasing could be a problem.

[_Wim_]

BG7TBL Noise source => DC-block => Stepped attenuator => CMU200 vs
BG7TBL Noise source => DC-block => Stepped attenuator => 50-ohm input adaptor => Picoscope input

As the same "drop" between 60Mhz and 160Mhz is seen as when I sweeped the CMU200 generator to the picoscope input (https://www.eevblog.com/forum/testgear/picoscope-hack/msg4545359/#msg4545359), I do start to believe the Pico is a little low between 60Mhz and 160Mhz...

[_Wim_]

Interesting, I would try with some other values for input resistor because with zero It would seem that aliasing could be a problem.

Might be worth a try, but I would have expected to see those aliasing effects in the spectrums measured above.

[markone]

Interesting, I would try with some other values for input resistor because with zero It would seem that aliasing could be a problem.

Might be worth a try, but I would have expected to see those aliasing effects in the spectrums measured above.

Do you have a sinusoidal signal generator capable to go over 200 MHz ?

[_Wim_]

Do you have a sinusoidal signal generator capable to go over 200 MHz ?

Yes, the CMU200 (can go to 2.7GHz), and you are correct, aliasing does occur for input signals > 300 Mhz (just did a quick test)

[_Wim_]

aliasing does occur for input signals > 300 Mhz (just did a quick test)

But even with the original 75 Ohm resistor, the aliased signal is only +-13dB lower (so a -20dBm 301Mhz Signal is aliased to 199Mhz 13dB lower as the input). Test ran in 12bit mode => 500MS/s

With the zero ohm resistor in place, we are -3dB only...

[_Wim_]

aliasing does occur for input signals > 300 Mhz (just did a quick test)

But even with the original 75 Ohm resistor, the aliased signal is only +-13dB lower (so a -20dBm 301Mhz Signal is aliased to 199Mhz 13dB lower as the input). Test ran in 12bit mode => 500MS/s

With the zero ohm resistor in place, we are -3dB only...

@egonotto, if you have RF generator that can send a 301MHz signal , would you be willing to run this also on your 5444B? I would like to know if this is an artifact from the "upgrade" or not. It could be that the LMH6574 is used to select between 4 anti-aliasing filters, and that the component values differ between the 5442B and the 5444B...

[egonotto]

Hello,

here are 3 pictures with sine with 50 Ohm termination 100 MHz 301 MHz and 320 MHz

Best regards
egonotto

[_Wim_]

Hello,

here are 3 pictures with sine with 50 Ohm termination 100 MHz 301 MHz and 320 MHz

Best regards
egonotto

Excellent! Thanks for confirming, this saves me from trying to fix something that would not have been fixable. 

[_Wim_]

A bit off topic, but not completely...

Because the BG7TBL noise source was very spiky between 0 and 2MHz, I had always been using a low cost DC block to act as a high pass filter. Today I made a quick and dirty filter board to get completely rid of these spikes (5th order filter), and also flatten the response up to 200Mhz.

After some experimentation, I achieved a flatness of < 2dBm between 10Mhz and 200MHz.

For reference, I attached also the spectrum measured up to 2000MHz measured with my CMU200, but the high frequency spectrum is not changed.

[jasonRF]

I can verify that there is at least one Picoscope that doesn't follow this hacking recipe.   

I have a 2204a I purchased in 2016.  The hardware version is 17.  Anyway, increasing byte 0B by 1 and decreasing byte 1B by 1 did not do the trick.  I have tried a bunch of other options as well, but so far all of the trials have resulted in a device that the picoscope software does not recognize. 

Edit:forgot to say that i am trying to turn it into a 2205a, which has identical hardware. 

The eeprom on my 2204a is definitely different than the one discussed on the sigroc page.  For example, B7-D9 and F8-FD are all non-zero.  I have of course mucked with some of these bytes as well - especially F8-FD that look less like cal data than some of the other areas. 

I am very inexperienced when it comes to this kind of stuff, so any suggestions you-all might have would be helpful.  For the trial and error approach I have been using so far, it might be time to write a python script to automate it more. 

In any case, it has been fun to play with this. 

cheers,

jason

[_Wim_]

I can verify that there is at least one Picoscope that doesn't follow this hacking recipe.   

I have a 2204a I purchased in 2016.  The hardware version is 17.  Anyway, increasing byte 0B by 1 and decreasing byte 1B by 1 did not do the trick.  I have tried a bunch of other options as well, but so far all of the trials have resulted in a device that the picoscope software does not recognize. 

Edit:forgot to say that i am trying to turn it into a 2205a, which has identical hardware. 

The eeprom on my 2204a is definitely different than the one discussed on the sigroc page.  For example, B7-D9 and F8-FD are all non-zero.  I have of course mucked with some of these bytes as well - especially F8-FD that look less like cal data than some of the other areas. 

I am very inexperienced when it comes to this kind of stuff, so any suggestions you-all might have would be helpful.  For the trial and error approach I have been using so far, it might be time to write a python script to automate it more. 

In any case, it has been fun to play with this.

cheers,

jason

Can you post the original bin file here?  Do you have somewhere a byte with the value "77" in the original binary?

[jasonRF]

I can verify that there is at least one Picoscope that doesn't follow this hacking recipe.   

I have a 2204a I purchased in 2016.  The hardware version is 17.  Anyway, increasing byte 0B by 1 and decreasing byte 1B by 1 did not do the trick.  I have tried a bunch of other options as well, but so far all of the trials have resulted in a device that the picoscope software does not recognize. 

Edit:forgot to say that i am trying to turn it into a 2205a, which has identical hardware. 

The eeprom on my 2204a is definitely different than the one discussed on the sigroc page.  For example, B7-D9 and F8-FD are all non-zero.  I have of course mucked with some of these bytes as well - especially F8-FD that look less like cal data than some of the other areas. 

I am very inexperienced when it comes to this kind of stuff, so any suggestions you-all might have would be helpful.  For the trial and error approach I have been using so far, it might be time to write a python script to automate it more. 

In any case, it has been fun to play with this.

cheers,

jason

Can you post the original bin file here?  Do you have somewhere a byte with the value "77" in the original binary?

Hi _Wim_,

Good question!  I did notice that byte F8 is 77 so is the nominal device type 119, and have played with increasing it by one and then decreasing the neighboring bytes (one at a time, of course) by one, but all to no avail.   I have attached a file with the eeprom dump I made using an arduino.  I wrote the file to include both hex and dec values because I don't think so well in hex. 

Edit: and if you want the file in a different format let me know.   I used a python script to talk to the arduino and dump this file, so is easy to do it again in a different format.

Thanks,

jason