Increasing logic analyzer input impedance

[michaeliv]

Hello,

I have a cheapo China Logic analyzer : http://www.aliexpress.com/item/-/2020974930.html
The input impedance is ~100kOhm.
I want to increase this to at least around 10MOhm.
The goal is to have it record pulses as accurately as it does natively.
I was thinking of either putting a high impedance op-amp configured as a buffer on each input, however I think to be able to keep the full 24MHz bandwidth , I would need op-amps with around 1000V/uS slew rate, which are quite expensive.
Would a N-MOSFET(BSS138) + work ? I'm thinking of using it in this configuration: https://upload.wikimedia.org/wikipedia/commons/thumb/4/4f/N-channel_JFET_common_source.svg/2000px-N-channel_JFET_common_source.svg.png
Any problems / timing issues you can predict? Any better solutions you can suggest  ?

Thanks!

Later Edit: There are 100k pull-ups on each pin, this causes the logic analyzer to output ground + power, acting like a power supply. I opened up the logic analyzer and converted the pull-ups to pull-downs. No major mods were required, only moving 1 resistor, 4 cut tracks, 2 bodge wires & solder bridge. Now all is OK, the logic analyzer only outputs ground, not ground + power, as the case was by default.

[Gyro]

Why do you want to increase the input RESISTANCE above 100k? It's a logic analyser and any logic is quite capable of driving 100k without problem.

The biggest obstacle to monitoring high speed logic signals is the CAPACITANCE of the input leads. The impedance caused by the lead capacitance of a cheap logic analyser is well below 100k. A mere 10pF of lead capacitance has an impedance of 1.5k at 10MHz - and that's for a sinewave. Your leads will be rather higher than 10pF!

Unless you have a way of lowering probe and lead capacitance there is absolutely no point in trying what you are proposing. Professional logic analysers have probe capacitance around 2pF but input resistance is still normally 100k.

[michaeliv]

Yes, Indeed I meant resistance.
The reason is that If I want to power cycle my DUT than I also have to power cycle the logic analyzer.
What happens is that my logic analyzer is outputting voltage on the pins ( to detect a ground connection I suppose ) and keeping some parts of my DUT powered up so that they don't completely loose state when disconnected from power.

[tggzzz]

The biggest obstacle to monitoring high speed logic signals is the CAPACITANCE of the input leads. The impedance caused by the lead capacitance of a cheap logic analyser is well below 100k. A mere 10pF of lead capacitance has an impedance of 1.5k at 10MHz - and that's for a sinewave. Your leads will be rather higher than 10pF!

Oh, don't forget the inductance! Rule of thumb is that 1mm->1nH. Calculate the resonant frequency of the relevant lead lengths and input capacitance; not pretty and easily visible. Also calculate the voltage over/undershoot during transitions.

[Gyro]

The input circuit of those logic analysers is very simple:

- There's a 100k pulldown resistor to ground, to define the logic state of an unconnected input. That's where your 100k input resistance is coming from. This could be increased or removed, but then the display trace for any unconnected probe would 'flap in the breeze'.

- Following the pulldown resistor there's a series resistor of lower value, probably around 1k to provide some degree of overload protection to the input logic.

- Finally there's the input logic, typically a 74HC245 (strapped as input only).

The logic analyzer isn't outputting any voltage on the pins - what happens is that when you power it down, is that the input protection diodes in the 74HC245 are pulling your signals down via the series resistors. With 1k series resistors and 0.7V protection diode drop that's about 4.3mA load on a 5V signal (less at 3v3 obviously).

You could try increasing the value of the internal series resistors to, say, 10k (they will be small smd resistor arrays), however the higher the value, the lower the frequency of the RC low-pass filter that you create with the HC245 input capacitance (approx 5pF).

Your easiest solution is probably to include an octal logic buffer on your circuit (another 245?) powered by the target that won't mind its outputs being loaded a bit, and use its inputs as the probes instead.

General practice though is to power down the target, or remove probes, before powering off test equipment.

[Gyro]

The biggest obstacle to monitoring high speed logic signals is the CAPACITANCE of the input leads. The impedance caused by the lead capacitance of a cheap logic analyser is well below 100k. A mere 10pF of lead capacitance has an impedance of 1.5k at 10MHz - and that's for a sinewave. Your leads will be rather higher than 10pF!

Oh, don't forget the inductance! Rule of thumb is that 1mm->1nH. Calculate the resonant frequency of the relevant lead lengths and input capacitance; not pretty and easily visible. Also calculate the voltage over/undershoot during transitions.

Thats why I keep my old Agilent E9340A Logicwave PC hosted analyzer around. It may only have been designed for Win98 (XP with DLL patch), it may only have 128k samples, and they may never have got around to implementing the USB port  but at least is uses the same 1.5pF passive probe technology as their mainframes and can do 'honest' sampling up to 250MHz.

These modern cheap analyzers and MSOs claiming 100MHz+ while using 'bits of wire' probes are a bit of a joke in that respect, useful as protocol analyzers and for low speed stuff at best.   Oops, controversial. 

[tggzzz]

These modern cheap analyzers and MSOs claiming 100MHz+ while using 'bits of wire' probes are a bit of a joke in that respect, useful as protocol analyzers and for low speed stuff at best.   Oops, controversial. 

Not to those With A Clue

[michaeliv]

The logic analyzer isn't outputting any voltage on the pins - what happens is that when you power it down, is that the input protection diodes in the 74HC245 are pulling your signals down via the series resistors.

Measuring it when idle - it outputs 3.3v and is able to deliver 35uA per channel. It seems to have 100k pull-up resistors on the output.
Will look into the 74HC245 solution and maybe MOSFET with pull-down as an option and see what works best.

[Gyro]

Ah, ok. Yours has pull-ups, not pull-downs. Same function though, to define the idle state of unconnected probes. Note that they will turn into pull-downs once the logic analyser is powered off - they will be 'attempting' to power the internal 3V3 supply (logic and micro).

[michaeliv]

Follow-Up for anyone facing the same situation, I opened up the logic analyzer and converted the pull-ups to pull-downs. No major mods were required, only moving 1 resistor, 4 cut tracks, 2 bodge wires & solder bridge. Now all is OK, the logic analyzer only outputs ground, not ground + power, as the case was by default.

[Gyro]

Glad you got it sorted the way you want. It's always nice when it comes out as the cheapest option too!

[alank2]

I read that the buffer used on these cheap clones is not 5V compliant.  Connecting it to a 5V source will allow 7mA to flow!  The solution (if you use 5V) was to replace the buffer with something like a MC74LCX245DTR2G.

[Gyro]

Thats why I keep my old Agilent E9340A Logicwave PC hosted analyzer around. It may only have been designed for Win98 (XP with DLL patch), it may only have 128k samples, and they may never have got around to implementing the USB port  but at least is uses the same 1.5pF passive probe technology as their mainframes and can do 'honest' sampling up to 250MHz.

These modern cheap analyzers and MSOs claiming 100MHz+ while using 'bits of wire' probes are a bit of a joke in that respect, useful as protocol analyzers and for low speed stuff at best.   Oops, controversial. 

EDIT: Necropost warning.

At the request of member jester42 via PM, I have attached the relevant DLL to allow the Agilent E9340A Logicwave to be used on Windows XP SP2. The zip contains the replacement DLL for an existing system one, together with a readme file.

Use at your own risk, virus check it, but it is a pretty old file now.

[jester42]

Thank you Gyro!! The driver worked! I now can use the E9340A again. I am forever in you debt. Jester

[Gyro]

You're welcome, glad it worked.