counting edges @ 90MHz

[Henrik_Volkers]

Hi

I'm looking for a solution to count edges from an external source without prescaler* @ 50 (90) MHz and read the counter (>=32bit) with a sampleclock of 100kHz.

The M/X series Daq have counters  but the external sources routings limit the counter source to about 25MHz 😞

* I want every edge!

[John_P1 · Accepted solution]

Hi Henrik,

Do you need 50 MHz or 90 MHz?

I suggest looking at the 6602, which uses a different digital front end that has a much higher bandwidth than any of the multifunction DAQ products.  50 MHz should be attainable without prescaling on the 6602.  90 MHz is pushing it and would depend on the signal source and quality of the connections.  I have in the past counted up to 100 MHz successfully with the board using one of NI's HSDIO devices as the clock source using an SMB to BNC connection.

One thing to note is that duplicate count prevention is enabled by default when performing buffered edge count tasks on the 6602.  You'll want to disable this via a DAQmx property node, otherwise you would be limited to 20 MHz (1/4 of the maximum on-board timebase).  The name "Duplicate Count Prevention" is a bit misleading, disabling it means that the count value is latched into the buffer on the edge of the counter source immediately following the clock signal (routed to the gate on the 6602).

Best Regards,

[Henrik_Volkers]

Hi John,

thank you for the fast response. Somehow I had in mind, that the external IO of the 660X was limited to 40MHz ... I was wrong 🙂

And thanks for the hint with the duplicate count prevention, I thought it's deactivated by default.

50/90: I have two possible applications, one require 50MHz counting, the other 90MHz.

I currently run a test with a frequency mixer to bring the signal down to 20MHz , so I can use a PXI-6251 I already have. It's not as nice as counting directly ... 

What frontend to the counter would best fit the RF needs I have ?

A 75Ohm coax cable directly to the connector?  (BTW do NI have an order number for that connector? Wasn't there a nice shielded box with connector to screw directly at PXI board for custom PCBs? )   

I currently have a comparator driver with 50 output impedance, how about running all at 50Ohm and use a matching resistor of 150 Ohm at the connection?

[John_P1]

Hi Henrik,

I think you're talking about the TB-2715, which is a screw terminal block that mounts directly to the PXI-6602.  I'd want to actually set this up before giving a recommendation--the TB-2715 might work out but I haven't tried it before at high frequencies.  I'm hesitant on recommending a screw terminal connector block because I have had issues with the SCB-68 passing through high frequency signals cleanly, but perhaps the 2715 doesn't have the same problems since it connects directly to the 6602.

In the past I have used the BNC-2121 cabled directly into the SMB clock out connector of a PXIe-6545.  Both the output impedance of the 6545 and the characteristic impedance of the SMB to BNC cable were 50 Ohms.  I actually did not terminate at the BNC-2121 as I found it wasn't necesary with the cable lengths that I was using (I don't recall exactly what they were).

The spec for the max external source for the 6602 only goes up to 80 MHz, so anything above that would not be a guarantee.  From past experience I believe it should work under the right circumstances, but it's probably something that would have to be set up to know for sure.  Do you have the contact info for your Field Sales Engineer who might be able to arrange for a loaner 6602 for you to try out?  Or, perhaps you can simply order one and return it if it does not suit your needs (check with your sales representative to confirm NI's return policy).

Best Regards,

[Henrik_Volkers]

---

@John P wrote:

Hi Henrik,

I think you're talking about the TB-2715, which is a screw terminal block that mounts directly to the PXI-6602.  I'd want to actually set this up before giving a recommendation--the TB-2715 might work out but I haven't tried it before at high frequencies.  I'm hesitant on recommending a screw terminal connector block because I have had issues with the SCB-68 passing through high frequency signals cleanly, but perhaps the 2715 doesn't have the same problems since it connects directly to the 6602.

Close 😉 I was thinking about that housing with the connector and a matching PCB core in Ultiboard for custom designs 🙂

In the past I have used the BNC-2121 cabled directly into the SMB clock out connector of a PXIe-6545.  Both the output impedance of the 6545 and the characteristic impedance of the SMB to BNC cable were 50 Ohms.  I actually did not terminate at the BNC-2121 as I found it wasn't necesary with the cable lengths that I was using (I don't recall exactly what they were).

The spec claims a 75Ohm input impedance for the 6602, well the mismatch probably don't hurt that much if you have short cables.

The spec for the max external source for the 6602 only goes up to 80 MHz, so anything above that would not be a guarantee.  From past experience I believe it should work under the right circumstances, but it's probably something that would have to be set up to know for sure.  Do you have the contact info for your Field Sales Engineer who might be able to arrange for a loaner 6602 for you to try out?  Or, perhaps you can simply order one and return it if it does not suit your needs (check with your sales representative to confirm NI's return policy).

I'm in close contact with NI here in Germany and a PXI-6602 was already offered for a test 🙂 * 

Best Regards,

---

A maybe root idea, but how about using a PXIbus breakout board and feed the 90MHz to one of the internal triggerlines? Would that drill a hole through the f-max limited input?

*

Spoiler

The PTB seems to be a good value customer 😉

[John_P1]

Hi Henrik,

From your last comment, it sounds like  you were considering connecting the 90 MHz over the PXI backplane to the PXI_Trig (RTSI)  lines of the 6602.  You can route signals to the counters from these lines, but the bandwidth would be less than the PFI lines on the front of the board.  In fact, the user manual explicitly points out that the RTSI bus cannot carry the 80 MHz timebase for bandwidth reasons (page 3-16).

The custom design sounds appealing, be sure to let us know what you end up going with if you come up with a working solution!

Best Regards,

[Henrik_Volkers]

Hi John,

I made some promising tests with a frequency mixer and brought it down to 20,  same can be done with the 90.

The mixer will subtract a known number of counts, the prescaler will throw away a fraction, reducing resolution.

So a simple USB M card will do the job :). 

OK...  this time, but I will monitor the counter market 😉

Spoiler

We have some guys here who would claim 20MHz 'unstable DC' 😄

[Henrik_Volkers]

---

@John P wrote:

From your last comment, it sounds like  you were considering connecting the 90 MHz over the PXI backplane to the PXI_Trig (RTSI)  lines of the 6602.  You can route signals to the counters from these lines, but the bandwidth would be less than the PFI lines on the front of the board.  In fact, the user manual explicitly points out that the RTSI bus cannot carry the 80 MHz timebase for bandwidth reasons (page 3-16).

---

I had in mind that there is a 100MHz Clockline  in the newer backplanes I might abuse .. 

However, while it's fun for me to mess around with analog circuitry, a (CPLD/VHDL) pipelined, sync, HS Counter design is not.

[John_P1]

There is a 100 MHz clock on the PXIe backplane as well as the DStar lines which would have sufficient bandwidth for your signal, but unfortunately the 6602 does not have access to them.  Interestingly enough, PXIe X Series cards do have access to these lines (see manual), so if you were intending on routing this signal through the backplane you would probably want to go back to X Series.  Getting the signal onto the backplane cleanly is another story however (the 6674T could do it I believe).

I'd still recommend trying the PFI lines of the 6602 going through a BNC terminal block.

Best Regards,