Equipment recommendation for PCB TDR analysis / S-parameter extraction

[dr_chernobyl]

I am doing some equipment research for my company for buying the instrument which would be capable of TDR analysis of relatively short PCB transmission lines and S-parameter extracion for creating the model of such lines. It would be good if there was also an option of doing eye diagram analysis.
 
 Basic specs that I was looking for was around ~20ps incident pulser risetime and 20-30GHz sampling head bandwidth. The caveat is it should be as cheap as possible, probably meaning that we are looking for equipment made in late 90s or early 00s at best. My budget would be around 20k USD maybe (and yes I know that this is a stretch for such high performance test equipment but my boss doesnt understand that clearly).
 
 So to get to the point, I found out about Tektronix DSA8200 mainframe and did some research which modules would be useful for us, for eg. 80E08 / 80E04 TDR modules seem ideal for my use case, and 80A05-10G clock recovery module would also be nice. My main question is what the gurus here think about this instrument and are there any better recommendations that are maybe cheaper or more useful? And what are some auxiliary pieces of equipment needed to run such a setup successfully?
 
 
 If you think its impossible to setup such system for ~20k USD, my main alternative is Tektronix 11801 series with couple of SD-24 sampling heads loaded. But this is a much older instrument that I think would be of limited usability and reliability problems based on age. Of course I appreciate any advice.
 
 Thanks in advance guys

[tszaboo]

I believe the new-ish way of doing TDR is with a VNA, not with a scope. Which is of course a completely different sort of instrument than what you had in mind, with it's own different challenges. But maybe take a look at those instruments, because some of them will do TDR, and the 1 port devices are not particularly expensive (Cheaper than a car) even at 18GHz.

[tggzzz]

Basic specs that I was looking for was around ~20ps incident pulser risetime and 20-30GHz sampling head bandwidth. The caveat is it should be as cheap as possible, ...

Those are mutually exclusive constraints Alternatively, define what "cheap" means in this context

Don't forget to include the cost of the cables and the calibration loads, including not exceeding the number of acceptable mating cycles

If your company is going to be relying on the results, don't forget to include whatever servicing/calibration is required by your customers.

Consider hiring the equipment.

[joeqsmith]

You can find the LeCroy SPARQs used on eBay for about that $$$.   Not sure about specs, export laws, tax.....   Just tossing it out there.

[dr_chernobyl]

Quote from: tszaboo on Today at 07:10:47

I believe the new-ish way of doing TDR is with a VNA, not with a scope. Which is of course a completely different sort of instrument than what you had in mind, with it's own different challenges. But maybe take a look at those instruments, because some of them will do TDR, and the 1 port devices are not particularly expensive (Cheaper than a car) even at 18GHz.

Well if I only needed to extract S-parameters and see how transmission lines behave in frequency domain I would probably be looking at buying some cheaper VNA, but to see where problems exactly are on the line, analyze crosstalk points and maybe do eye diagram analysis I think sampling scope is the best bet for me. There are some high end Siglent VNAs that have TDR option available but they also cost pretty penny :/



Quote from: tggzzz on Today at 08:41:35Quote from: dr_chernobyl on Yesterday at 23:38:58

Basic specs that I was looking for was around ~20ps incident pulser risetime and 20-30GHz sampling head bandwidth. The caveat is it should be as cheap as possible, ...

Those are mutually exclusive constraints Alternatively, define what "cheap" means in this context

Don't forget to include the cost of the cables and the calibration loads, including not exceeding the number of acceptable mating cycles

If your company is going to be relying on the results, don't forget to include whatever servicing/calibration is required by your customers.

Consider hiring the equipment.

Yeah I know, but explain that to the management
I understand their philosophy that we need to do things as cheap as possible and rely on simulations because in their mind simulations on modern software are flawless but I feel blind without actually measuring stuff from time to time and doing some experimental verification of simulation results....
And hiring is not an option because my team lead and I would like to set up our own small lab for testing, we tend to need the results ASAP, renting is a bit of a hassle.

DSA8200 really caught my eye because of so much capability in one instrument, and my definition of cheap in this context is getting best bang for my buck. Older instruments may be cheap but they once were the best you could buy, and that would be enough for me.

[tggzzz]

Is it possible to create an explanation that is acceptable to management?

Concepts like...
time to do it without equipment due to repeated attempts to get it right...
time to do it with equipment...
cost of hiring equipment...
market window closes after time...

One company I worked at justified the cost of a hardware simulator (~50 years salary) on the grounds it got the product to market 3 months sooner.

Don't know current figures, but an old rule of thumb was that hiring equipment for 10 months cost the same as buying it.

[tszaboo]

Quote from: tszaboo on Today at 07:10:47

I believe the new-ish way of doing TDR is with a VNA, not with a scope. Which is of course a completely different sort of instrument than what you had in mind, with it's own different challenges. But maybe take a look at those instruments, because some of them will do TDR, and the 1 port devices are not particularly expensive (Cheaper than a car) even at 18GHz.

Well if I only needed to extract S-parameters and see how transmission lines behave in frequency domain I would probably be looking at buying some cheaper VNA, but to see where problems exactly are on the line, analyze crosstalk points and maybe do eye diagram analysis I think sampling scope is the best bet for me. There are some high end Siglent VNAs that have TDR option available but they also cost pretty penny :/



Quote from: tggzzz on Today at 08:41:35Quote from: dr_chernobyl on Yesterday at 23:38:58

Basic specs that I was looking for was around ~20ps incident pulser risetime and 20-30GHz sampling head bandwidth. The caveat is it should be as cheap as possible, ...

Those are mutually exclusive constraints Alternatively, define what "cheap" means in this context

Don't forget to include the cost of the cables and the calibration loads, including not exceeding the number of acceptable mating cycles

If your company is going to be relying on the results, don't forget to include whatever servicing/calibration is required by your customers.

Consider hiring the equipment.

Yeah I know, but explain that to the management
I understand their philosophy that we need to do things as cheap as possible and rely on simulations because in their mind simulations on modern software are flawless but I feel blind without actually measuring stuff from time to time and doing some experimental verification of simulation results....
And hiring is not an option because my team lead and I would like to set up our own small lab for testing, we tend to need the results ASAP, renting is a bit of a hassle.

DSA8200 really caught my eye because of so much capability in one instrument, and my definition of cheap in this context is getting best bang for my buck. Older instruments may be cheap but they once were the best you could buy, and that would be enough for me.

I'm not up to date which VNA has TDR as an option, or by default. The one I'm using (Megiq) goes to 4GHz, and that has TDR. If you have a hard time convincing management about the costs of buying, then try rental. If you also have the software simulation, and you can show the difference in results of our imperfect but beautiful world, compared to the simulation, they might spend the money.
If you actually need the equipment for what you are making.

[DaJMasta]

You can find the LeCroy SPARQs used on eBay for about that $$$.   Not sure about specs, export laws, tax.....   Just tossing it out there.

Was my first thought, too.  Marketed as low budget, high bandwidth VNAs, they're multiport TDR devices from the 2000s that seem to be designed for similar types of measurements (I think some promotional material deals with PCB transmission lines specifically.)  They certainly wouldn't break the budget, but your test fixtures might...

[Bud]

I am doing some equipment research for my company for buying the instrument which would be capable of TDR analysis of relatively short PCB transmission lines...

How short ?

[dr_chernobyl]

I am doing some equipment research for my company for buying the instrument which would be capable of TDR analysis of relatively short PCB transmission lines...

How short ?

for example we would like resolution of about 1-1.5mm on PCB to see what is exactly happenning on each segment, and as I understand you need pretty fast incident risetime for such resolutions

Ive been looking at SPARQ analyzer, and for that kind of money it seems like a great deal, datasheet says 6ps pulser rise time and 40G of BW for 4004E model which is around 20k USD

[tggzzz]

I am doing some equipment research for my company for buying the instrument which would be capable of TDR analysis of relatively short PCB transmission lines...

How short ?

for example we would like resolution of about 1-1.5mm on PCB to see what is exactly happenning on each segment, and as I understand you need pretty fast incident risetime for such resolutions

Fun

I strongly suggest you find out what risetime/bandwidth will be necessary for that. If nowhere else, start with Bogotin's rules of thumb.

There's also the simple point that if it can be determined your system "won't care" about such short lengths, then you needn't measure them. Such advance determination would be sound practical engineering. Again, Bogotin might provide a starting point.