DRS4 Forum
  DRS4 Discussion Forum, Page 36 of 45  Not logged in ELOG logo
New entries since:Thu Jan 1 01:00:00 1970
ID Date Authorup Subject
  707   Fri Jun 29 07:51:33 2018 Stefan RittNegative Bin Width

Yes that's normal. A negative cell bin width means that the next cell N+1 samples the input signal before cell N. This can happen due to the signal routing on the DRS4 chip.

Stefan

Woon-Seng Choong wrote:

I am using a DRS4 Evaluation Board v5 and running the drsosc.exe version 5.06 on a Window 7 machine. I have performed the voltage and timing calibration.

With test pulses on channel 1 and 2, I collected binary data file with all 4 channels active sampling at 5GSPS. 

Attached is a distribution of the bin_width vs. cell # for all the 4 channels. Note that there are few cells with bin_width < 10 ps. 

Channel 1: bin_width[498] = -0.000348, bin_width[1010]= -0.000348

Channel 2: bin_width[498] = 0.007363

Channel 3: bin_width[498] = 0.007843

Channel 4: bin_width[498] = 0.005948

Is this normal? How can you get negative bin_width? What does negative bin_width means?

I have attached the binary data file for your verification.

 

 

  712   Tue Aug 14 06:10:49 2018 Stefan RittLatch delay support

I put that on the wish list, but I won't have time for that in the next months.

Stefan

Martin Petriska wrote:

Hi,

https://forge.physik.rwth-aachen.de/projects/drs4-rwth

Not sure about their licensing, but is it possible to add latch delay support to official firmware ?

Best regards

Martin

 

  713   Tue Aug 21 14:36:44 2018 Stefan RittOptimal readout speed

The analog output of the DRS4 chip needs some time to settle. In principle it need an infinite amout of time (exponential curve) to settle to 100% of the final value. So if we sample after a finite time, there is some error we do. Some of the error will be taken care of the voltage calibration, but there remains some residual error depending on the value of the previous sampling cell. So all sampling speeds 10 MHz, 16 MHz, 33 MHz are kind of rule of thumbs. In the end we run the evaluation board at 16 MHz to save a little bit of power (which is limited on an USB device). But I never made a careful study of noise-after-calibration vs. sampling speed. If you have some measurements, I'm happt to include it in the data sheet.

Stefan

Sean Quinn wrote:

Dear DRS4 team,

On page 3 of the data sheet, Table 1. for readout speed a typical value of 10 MHz is specified, but in the comment column it notes optimal performance achieved at 33 MHz.

I see the V5.1 eval board runs at 16 MHz. I'd like to understand the rationale for this using speed, instead of 33 MHz. Is there an SNR issue for the ADC at the higher speed, even though this is optimal for the DRS4?

Very best,

Sean

 

  715   Tue Sep 4 13:04:30 2018 Stefan Ritt"Symmetric spikes" fixed

Yes it's possible, but I have to find time for that. The software of the evaluation board takes care of the spikes ("remove spikes"), so I thought it's not so urgent to fix that in the FPGA (which takes me some time).

Stefan

Martin Petriska wrote:

Hi,

Is it possible to fix it by FPGA changes?  I see readout cycle (proc_drs_reedout) in drs4_eval(4)5_app.vhd, but not sure where to exactly put this three commands. Could you please attach app.vhd file for eval board with example how to fix ?

Regards,

Martin

 

Stefan Ritt wrote:

Good news for all DRS4 users. After many years, I finally understand where the "symmetric spikes" come from and how to fix them.

The "symmetric spikes" are small spikes of 17-18mV, which randomly happen at 1-2 cells. They alwas come in groups of 2 in each channel, symmetric around sampling cell #512. See first attachment.

The reason for the spikes is the previous readout cycle. On each readout cycle, the "read bit" is clocked through all 1024 cells to switch one cell contents to the DRS4 output. At the end of the 1024 cycles, the read bit stays at its last position. The bit is carried by a metal line on the chip, which crosses all 9 channels (second attachment). This bit now influences the sampling cells below the metal line capacitively, so their contents is "pushed up" by a few mV, just like the ROFS offset does. Since the DRS sampling channels are in a snake layout, going 0-512 from left, then 512-1023 back again, the line crosses two cells in each channel, and thus the symmetric spikes.

Previously, there was a software solution for that. In the evaluation board software DRSOsc there is a button "Remove spikes" which tries to fix this in software. Although this works most of the time, it's annoying and not 100% safe. Like when the spike sits on top of a noise signal, it might not be recognized. Fixing this in hardware is however straight forwar. After the readout cycle ends, push the read bit out of the chip:

  • Address the read shift register by applying 1011b to A3:A0
  • Switch SRIN low
  • Apply 1024 clock cycles to SRCLK

This shifts the bit out of the chip, so that the next event is not affected by the read bit. The third attachment show the effect of this. The "clear cycle" increases the readout time a little bit, but depending on the application this might be worth it.

Regards,
Stefan

 

 

  718   Wed Sep 26 14:44:14 2018 Stefan RittTrigger OUT pulse width variable from 100 us up to 100 ms

The "Trigger OUT" has changed recently. It goes high on a new trigger, but then STAYS high until the board has been read out by the PC and re-started. This allows better synchronization with some external trigger, which can be re-armed with the falling edge of the trigger out signal. The signal can be quite long, since readout of an event via USB typically takes 2 ms, but can be more if the PC is busy. If you need back your 150 ns pulse, send the trigger out to an external pulse shaper with fixed shaping width.

Stefan

Gerard Arino-Estrada wrote:

Hello Stefan,

I am using the DRS4 board connected to a Raspberry PI and through the drsosc application. I am interested on using the "Trigger OUT" signal to do some extra data processing with NIM modules. According to the manual, for each hardware trigger a TTL pulse of 150 ns width should be send through the "trigger OUT". In my case I do see pulses with widths ranging from 100 microseconds up to hundreds of miliseconds. I am connecting the signal directly to an oscilloscope with 50 Ohm termination. I have tried two DRS4 boards in identical conditions and both show the same behavior.  Having such wide and variable pulses makes it complicated for me to do the extra post-processing. Have you any idea of what might be going wrong? Thank you very much.

Best regards,

Gerard

 

  721   Wed Sep 26 19:21:03 2018 Stefan RittTrigger OUT pulse width variable from 100 us up to 100 ms

In meantime I even updated the manual.

Stefan

Gerard Arino-Estrada wrote:

Thank you very much for the answer, I really appreciate your help.

Thanks!

Gerard

 

  723   Thu Nov 8 09:57:26 2018 Stefan RittPi attenuator on eval board inputs?

The attenuator compensates for the gain of the buffer which is slightly above one. In addition, it serves as a "placeholder" in case one wants larger input signals. One can easily convert the attenuator to -6db, -12db, etc. by chaning the resistors.

Stefan

Sean Quinn wrote:

Dear DRS4 team,

 

I am curious about this part of the circuit:

What is the purpose of this?

 

  725   Thu Nov 8 11:54:33 2018 Stefan RittTiming Issue
That's not a bug, but a feature of the DRS4 chip. The time bins have different values by the properties of the chip. They are generated by a chain of inverters, which all have different propagation times. This delay is measured by the time calibration and then applied. If you want equidistant bins, 
you have to interpolate your data points (linearly or by splines) and resample the signal. You can find more details in the DRS4 data sheet.

Best,
Stefan


> Hi,
> 
> We are using the DRS4 Evaluation Board as a digitizer in our laboratory.
> We found a strange behavior in the saved files, more specifically the time difference between two consecutive points is not constant, also after the Timing Calibration.
> As an example, I paste a piece of a xml file saved using the drsosc program, acquiring CH1 (open):
> 	
> 	---------------------------
> 	---[ START XML EXAMPLE ]---
> 	---------------------------
> 	
> 	<?xml version="1.0" encoding="ISO-8859-1"?>
> 	<!-- created by MXML on Thu Nov  8 11:13:27 2018 -->
> 	<DRSOSC>
> 	<Event>
> 	<Serial>1</Serial>
> 	<Time>2018/11/08 11:13:27.163</Time>
> 	<HUnit>ns</HUnit>
> 	<VUnit>mV</VUnit>
> 	<Board_2796>
> 	<Trigger_Cell>216</Trigger_Cell>
> 	<Scaler0>0</Scaler0>
> 	<CHN1>
> 	<Data>0.000,-1.0</Data>
> 	<Data>1.083,-1.0</Data>
> 	<Data>2.143,-1.0</Data>
> 	<Data>2.926,-1.0</Data>
> 	<Data>4.249,-0.1</Data>
> 	<Data>4.929,-0.6</Data>
> 	<Data>6.075,-0.4</Data>
> 	<Data>7.042,0.0</Data>
> 	<Data>8.299,0.2</Data>
> 	
> 	[...]
> 	
> 	-------------------------
> 	---[ END XML EXAMPLE ]---
> 	-------------------------
> 
> We found the same behavior saving events in the binary format, and then reading them with the read_binary.cpp
> 
> Is there a way to fix our issue?
> 
> Thanks a lot
> 
> Davide and Alessio
  729   Wed Jan 30 08:02:25 2019 Stefan RittDRS4 domino wave stability study

The Domino wave is most stable at 5 GSPS, slowly degrades down to 3-2 GSPS, and at 1GSPS gets some significant jitter. This is for internal reasons in the chip and cannot be compensated by the loop filter. It is therefore important to run it as fast as possible if you want to achieve best timing resolution. As a rule of thumb, the jitter at 5 GSPS is about 20-25 ps, and at 1 GSPS it is maybe 150 ps. If you require good timing resolution, you can use the 9th channel to sample a stable reference clock (100 MHz for example) and measure timing relative to that clock. This way you can bring down the resolution to a few ps at 5GSPS and to maybe 40 ps at 1 GSPS.

Stefan

Saurabh Neema wrote:

We have been using DRS4 IC in our design for quite some time and it is giving good performance.

Till now we were using Domino wave frequency as 1 GSPS by use of reference clock to DRS4 and internal PLL of DRS4. Recently we tried to use 4GSPS by modifying the reference clock.

What I have found that DRS4 domino wave is more stable at 4 GSPS as compared to 1 GSPS by doing the timing jitter analysis. I am not sure if it is the property of DRS4 IC to be having more stable domino wave at higher frequency (by design) or it is due to some external effects like PLL loop filter or any other on board parasitic effects.

Please share if anyone has done any study of DRS4 Domino wave stability at different sampling frequencies.

Thanks,

 

 

  730   Wed Jan 30 17:08:58 2019 Stefan RittROOT Macro for data acquired with the newest software

This one elog:361 should still work.

Stefan

Abaz Kryemadhi wrote:

Hello,

Is there a root macro for decoding binary data acquired with the newest software for single board or multi-boards daisy chained?

Cheers,

Abaz

 

  732   Sat Feb 2 10:10:22 2019 Stefan RittSaving Rate (only 15Acq/s)

The reduction of rate is because you save in XML format, which is an ASCII format, so human readable, but takes long to write. If you switch to binary format and write on a decent fast hard disk, you should get back to 450 Acq/s.

Stefan

Hans Steiger wrote:

Dear All,

 

when I use my Evaluation Board with some PMTs I can digitize 450 Acq/s or so. But when I want to save the waveforms the rate goes down. The Acqu. rate with saving is in the range of 14Hz up to 24 Hz.

I normally use the .txt file. I try to use the 5GS/s but also with much lower sampling rate the saving rate is not getting much better. 

Is this a problem of my McBook connected to the Evaluation Board?

 

All the best,

 

Hans 

 

  734   Mon Feb 4 16:46:04 2019 Stefan RittDifferent Distances between the sampling points

The sampling points are NOT equidestant, they have varying bin widths of 150ps to 250ps at 5GS/s. That's due the way the DRS4 chip works. You might have neglected that fact in the past, but that would have led to poor timing resolutions (typically 1-2ns resolution only). To get bins with the same width, you have to treat your waveform as a real X/Y points (or better U/T), and the re-sample that cure, maybe spline-interpolated, at 200ps bins.

Stefan

Hans Steiger wrote:

Dear All,

with the older software for my V5 Board i did not have the problem, that the distance between the sampling points (in time) is not the same (e.g. a sampling point all 200ps for 5GS/s). 

How can i fix this?

Can someone provide me the software for the board which is old enough to not have this problem. All my Root interpreters produce problems with this new data format.  Which version would be old enough?

 

All the best and thanks a lot,

 

Hans

 

  736   Mon Feb 4 18:18:22 2019 Stefan RittDifferent Distances between the sampling points

 elog:361

Hans Steiger wrote:

Sorry.... but is there a solution or a Root Macro, that reads the waveforms into a Root-Tree? I simply can not work anymore with the data. 

Can you tell me, which software was in use in early 2017?

All the best,

 

Hans

 

Stefan Ritt wrote:

The sampling points are NOT equidestant, they have varying bin widths of 150ps to 250ps at 5GS/s. That's due the way the DRS4 chip works. You might have neglected that fact in the past, but that would have led to poor timing resolutions (typically 1-2ns resolution only). To get bins with the same width, you have to treat your waveform as a real X/Y points (or better U/T), and the re-sample that cure, maybe spline-interpolated, at 200ps bins.

Stefan

Hans Steiger wrote:

Dear All,

with the older software for my V5 Board i did not have the problem, that the distance between the sampling points (in time) is not the same (e.g. a sampling point all 200ps for 5GS/s). 

How can i fix this?

Can someone provide me the software for the board which is old enough to not have this problem. All my Root interpreters produce problems with this new data format.  Which version would be old enough?

 

All the best and thanks a lot,

 

Hans

 

 

 

  738   Wed Feb 20 08:08:42 2019 Stefan Rittmeg?

You have to change the path to libusb-1.0.lib to the one where you installed it.

Stefan

Lev Pavlov wrote:

Hey. Strange problem. Why does the compiler refer there at all? Library installed drsosc works

 

LINK : fatal error LNK1104: cannot open file "C:\meg\online\drivers\drs\libusb-1.0\libusb-1.0.lib"

 

  740   Wed Feb 20 12:56:56 2019 Stefan Rittmeg?

No idea. Maye some access problem. Have you tried to start your program under an admin account?

Stefan

Lev Pavlov wrote:

Great, drs_exam compiles without problems. Now when you run the compiled file drs_exam writes board not found, but drsosc and drscl work without problems. What could possibly be the matter?

thanks for your patience

Lev

Stefan Ritt wrote:

You have to change the path to libusb-1.0.lib to the one where you installed it.

Stefan

Lev Pavlov wrote:

Hey. Strange problem. Why does the compiler refer there at all? Library installed drsosc works

 

LINK : fatal error LNK1104: cannot open file "C:\meg\online\drivers\drs\libusb-1.0\libusb-1.0.lib"

 

 

 

  743   Thu Feb 21 09:57:53 2019 Stefan Rittno board found

Could be. Have you tried that elog:657

Stefan

Lev Pavlov wrote:

Hey. Yes, the program is running as administrator. By the way, this is win10. Your drs_exam works fine. My drs_exam compiled wrote no board found. Maybe this is a problem like in the post https://elog.psi.ch/elogs/DRS4+Forum/698. Maybe there were solutions to the problems?

Thank You

Lev

Stefan Ritt wrote:

No idea. Maye some access problem. Have you tried to start your program under an admin account?

Stefan

Lev Pavlov wrote:

Great, drs_exam compiles without problems. Now when you run the compiled file drs_exam writes board not found, but drsosc and drscl work without problems. What could possibly be the matter?

thanks for your patience

Lev

Stefan Ritt wrote:

You have to change the path to libusb-1.0.lib to the one where you installed it.

Stefan

Lev Pavlov wrote:

Hey. Strange problem. Why does the compiler refer there at all? Library installed drsosc works

 

LINK : fatal error LNK1104: cannot open file "C:\meg\online\drivers\drs\libusb-1.0\libusb-1.0.lib"

 

 

 

 

 

  745   Mon Feb 25 08:48:27 2019 Stefan Rittno board found

"dynamic" or "static" does not matter, as long as you don't use your program on another computer. I have no more idea about the "no board found" problem. It works ok on all computers I tried at our lab.

Stefan

Lev Pavlov wrote:

 

Hello. When compiling drs_exam, do you need to use a "static "version of usblib or a "dynamic" version?"The problem with "no board found" is not solved. Thanks for your help.

Lev

  750   Fri Apr 12 09:55:50 2019 Stefan Rittmulti-board

Subtract 16 ns from your measured value ;-)

Stefan

Lev Pavlov wrote:

Good afternoon, I use 5 boards in multi-mode, everything is connected according to the instructions. Can I measure the phase difference between the two signals on channel 1 and channel 20? with each board the phase shift is added +16 ns I can not figure out how to compensate for this. give thanks

 

  752   Fri Apr 12 12:50:18 2019 Stefan Rittmulti-board

If you have two signal going through two cables, the cable have never the same length (on a scale of picoseconds), and you have to calibrate that anyway. So a proper timing calibration is not a crutch.

What do you mean by "manual 50ps"? The manual does not mention any resolution. In my experience, you can achieve about 10ps between channels of the SAME board easily. The phase shift between boards in multi-mode is always there, unfortunately there are no cable which conduct current faster than the speed of light! What you can do is to split a common reference clock and send a copy to one channel of each board, then calculate the timing relative to the next edge in that reference signal. This way you get rid of the phase shift, but this is also a kind of calibration, so in your laguange that would be "a big crutch".

Stefan

 

Lev Pavlov wrote:

 

I understand this, thanks. But my Chief does not understand this, he wants to see the phase difference without “crutches”. And what is meant in the manual 50 ps resolution? Maybe I just do not understand something? And if you submit a reference signal not in the mode of a garland, but simultaneously in parallel to all the boards, will this shift go? Thanks

Lev Pavlov

Stefan Ritt wrote:

Subtract 16 ns from your measured value ;-)

Stefan

Lev Pavlov wrote:

Good afternoon, I use 5 boards in multi-mode, everything is connected according to the instructions. Can I measure the phase difference between the two signals on channel 1 and channel 20? with each board the phase shift is added +16 ns I can not figure out how to compensate for this. give thanks

 

 

 

  754   Fri Jun 21 12:54:47 2019 Stefan RittEvaluation firmware wait_vdd state

Dear Andrew,

the posting you mention is still accurate. Any power supply will drop when you start the Domino wave, no matter how big your capacitor is. Unfortunately the output signal of the DRS4 scales with VDD. So if your VDD drops by 40 mV and you get a trigger and you immediately start the readout, the output baseline will also be shifted by about 40 mV. If you are sensitive to dead time, you can remove the wait_vdd state completely, but then you have to deal with varying baseline shifts. If you have narrow signals sitting on a broad baseline, you can correct for this by measuring the baseline outside your signal, then subtracting it before integrating your pulse. If you have lots of pile-up in your signals, it might sometimes be hard to evaluate the baseline on an event-by-event basis.

Stefan

Andrew Peck wrote:

Dear Stefan,

I am working with others at UCLA on a custom made board built around the DRS4. We are in the process of writing firmware so I am adapting the readout state machine from the evaluation board firmware.

I see in the state machine of the eval board firmware that after a trigger is received, the FPGA goes into the start readout state and then into "wait_vdd", where the FPGA waits "~120 us for vdd to stabilize" before reading out the ADC.

Our application is sensitive to deadtime and this wait_vdd state adds very significantly.  I am trying to find anything explaining the necessity of wait_vdd in the documentation / elog and have only found so far your old forum posting, https://elog.psi.ch/elogs/DRS4+Forum/12

Does this forum posting explain wait_vdd or is there a another purpose that I have missed?

If this post is relevant to wait_vdd, does the advice of large capacitance and an LDO with fast transient response still apply or are there any new recommendations?

Thank you,

Andrew Peck

 

ELOG V3.1.5-fe60aaf