Transmission Line Avalanche Pulse Generator

[Jay_Diddy_B]

Hi Group,

I would like to share some details of a Transmission Line Avalanche Pulse generator that I built.

The challenge is to try and generate a flat-topped waveform.

Here is a 4ns pulse being displayed on a Tektronix 7104 1GHz oscilloscope.


The rest of the story is in the pdf document attached.

Jay_Diddy_B

[muvideo]

Nice, thank you for sharing, it's always fascinating
the exploration of such high speed events.

I had success using a STC945, but since I'm cheap,
I built a simple discrete step-up for the power supply

The pulse profile I obtained was similar to yours,
but slightly longer at 10nS total, I was limited by
my plugin speed:
http://www.flickr.com/photos/14336723@N08/7248441944/#

Fabio.

[free_electron]

Good show old man ! ( quote from the addams family movie )

As i need to send out the PCB for the JW pulser this weekend i will adapt it so it has a transmission line on it. don;t know if it will work... i already had the layout adapted so it only uses traces ont he top. so i can use the bottom side of the board to do the wave guide.

i recalculated and will use 25 mil track with 8 mil gap . don't know how long i will be able to make the track ... i'll do the layout so that you can solder the 1 meg resistor into te classic layout or solder it into the waveguide and 'dot' the other end ento the ring around the collector

(see the JW pulse generator topic here)

[CarlG]

Nice work & good report!

[BravoV]

Thanks for sharing this, and what a bunch of nice scopes you got there ! 

[Mechatrommer]

if no abberation there, the rise time should be around 400ps guesstimating from my eyeball.
edit: i agree to prolonge the ON period to greater than 10ns minimum, so there's no dispute on rise time for slower DSO. at 4ns pulse, a 100MHz DSO will not reach max volt reading by the time the pulse is going down.

[alm]

Which is about right for a 1 GHz scope. You would need a sampling scope to properly characterize this circuit.

[tinhead]

i did similar efforts, finally ended up with 2m of RG316, based on AN94 version http://www.linear.com/docs/4183

My results are so far ok, you can compare to your TDS700 pics to see the difference.

EDIT: i'm using Mini Circuits 20db attenuator between the pulser and scope to optimize the damping.

[Jay_Diddy_B]

Hi,
I am working on a new version. The new version will have a pre-trigger so that I can use my Tektronix SD-24 sampling head in an 11801A  mainframe. This is a 20GHz scope. The sampling scope requires a pre-trigger.

I have an attenuator built in to my board.

I hope to have some results later today.

Stay tuned ....

Jay_Diddy_B

[Jay_Diddy_B]

Hi,
I am going to share some preliminary pictures and test results. I will take some better scope shots when I am finished.

The results were measured with a BFR505 transistor. The BFR505 will avalanche with around 30V.

The scope shots were taken on a Tektronix 11801 scope with a SD-24 sampling head. This combination has a bandwidth of 20GHz and a risetime of 17.5 ps 

Blank Board component side:


Blank Board solder side:



Rise time timebase 200ps / div. Measured risetime is around 100ps



Full pulse at 2ns / div



It will be interesting to see if the waveform cleans up when I put in all the 'Z' wires. Have left them out because I want to experiment with them.


Photograph of the assembled board, all the important stuff is the lower left corner.


Photograph showing the SMD transistor



Component side. The 74HC14s are used to generate the pre-trigger timing.



I will share more details later.

Jay_Diddy_B

[Jay_Diddy_B]

Hi Group,

I have done a little more work. I have installed about 1/2 of the 'Z' wires that connect the top and bottom ground planes. I have also taken some scope shots with some other scopes.

Tektronix TDS 754A 500MHz Bandwidth


Tektronix DSA602A with 11A71 Plugin 1 GHz


These were taken with a Tektronix 11801 and SD-24 sampling system. 20 GHz BW , 17.5ps RT.








I have also attached the schematic in pdf format.

Jay_Diddy_B

[tinhead]

I have also taken some scope shots with some other scopes.

Tektronix TDS 754A 500MHz Bandwidth

mixed feelings ... yours looks like Figure 11 in AN94, where my compared looks like Figure 13. The rise time difference (71ps) is "huge", these scopes are normally good up to ~600MHz, these 735ps are in my opinion not good enough.

[mazurov]

JDB,

I can't find a ferrite bead noted on a schematic. I'm trying to google 'we74279203'. Could you please share the manufacturer/model?

Thank you,
Oleg.

[tinhead]

WE is "Wurth Elektronik", simply search for number

[Jay_Diddy_B]

Hi,

for mazurov:

The Ferrite bead is from Wurth Electronik. Some of these parts are available from Digikey. This part is very similar

Digikey 732-1602-1-ND

The bead is not super critical.

for tinhead:

My risetime has not been peaked. I don't have a capacitor on the collector of the transistor, only the transmission line. It opens up a much bigger question as to why the risetime on an oscilloscope looks the way it is.

If a scope had a single pole at the bandwidth frequency then the waveform would look exponential.

Oscilloscopes have been engineered to have a Gaussian Response.

Here is a SPICE model of a Gaussian Filter




The reponse in the Frequency Domain




The response in the Time Domain




One of the properties of the Gaussian filter is that as you reduce the risetime on the input the risetime on the output reaches a minimum. After this point the risetime on the output becomes fixed.

HP made some Transistion Time limiters based on this property. The model numbers are 15432 through 15435

Link http://www.home.agilent.com/en/pc-1000000488%3Aepsg%3Apgr/transition-time-converters?nid=-536902323.0.00&cc=CA&lc=eng


What is interesting about this, is that the waveform from my pulse generator looks flater on a low bandwidth scope than it does on a higher bandwidth model.

The bottom line is that you should not see overshoot unless the input is overshooting.

Is there a Gaussian Conspiracy ? 

Jay_Diddy_B

[jahonen]

Hi,

Here is a SPICE model of a Gaussian Filter



Jay_Diddy_B

Isn't a Gaussian filter supposed to be like cascaded RC instead of cascaded LC?

Regards,
Janne

[Jay_Diddy_B]

Hi,

This application note from Picosecond Pulse Labs discusses the Gaussian Filter.

http://www.picosecond.com/objects/AN-7a.pdf

By strict definition the filter I implemented is 'near Gaussian' I modelled the G10 filter as described in the Picosecond app note.

Jay_Diddy_B

[tinhead]

Oscilloscopes have been engineered to have a Gaussian Response.

The rise time of this pulse generator is in ps range, this is where a typical digital scope respond flat and not gaussian anymore

http://cp.literature.agilent.com/litweb/pdf/5988-8008EN.pdf

therefore, and this is my opinion and why i've set damping in my generator as i did, it does not make sense to "cut" the signal to match gaussian response.

[Jay_Diddy_B]

Hi,
 It is a very interesting application note from Agilent. It basically say that oscilloscopes fall into two groups Gaussian and Flat response. They present the argument that the Flat Response instruments will give more accurate risetime measurements.

The essential difference is that in the Flat Response instruments the analog front end does not restrict the bandwidth, and an anti-aliasing filter dominates the response of the instrument.

The application note references Infiniium 58400 series of oscilloscopes. So I did a little digging into the specifications. This is what I found:



This suggests that only the top of the line 2.5GHz 54846A is flat response. All the other members are Gaussian.

I also looked at the Tektronix TDS520 manual. I believe that all the 500MHz TDS scopes have similar performance, but I might be wrong. And this is what I found there:





These specifications and the second formula provided by Tektronix suggest that the Tektronix TDS 500 series are in fact Gaussian.

It is also interesting to note that the risetimes are not measured directly, but calculated from the bandwidth measurement.


Jay_Diddy_B

[tinhead]

These specifications and the second formula provided by Tektronix suggest that the Tektronix TDS 500 series are in fact Gaussian.

right, TDS500/600/700 belongs for sure to Gaussian response group, the newer series TDS2000/TDS3000 as well.
TDS7000 should be however Flat response model.

I did some measurments of TDS700(with 250MHz LPF enabled) and Tekway TDS1202B -> this is a chinese DSO, with 200MHz BW. It seems to be as well in flat group, from the values it does match Butterworth. The same input stage is being used in lot of chinese DSOs, another example is Rigol DS1102e (all these Rigol hardware clones made by ATEN and Siglent).



It could be insteressting to see, properly measured, frequency respond of Rigol DS1102E.

[PA4TIM]

http://www.pa4tim.nl/wp-content/uploads/2010/11/calibratiescoop.png
Picture of bandwidth test using a sinewave. The Rigol DS1102E is included.

[Mechatrommer]

can you post it at https://www.eevblog.com/forum/reviews/frequency-response-of-your-dso/ including the excel file? whats the relation between dBmV, mVtt, mV rms and dB columns? noob question here i hope you can reply in that thread, thanks.

[Jay_Diddy_B]

Hi Group,

I have been busy adjusting my Transmission Line Avalanche Pulse Generator. Here are the latest results:

Picture 1 - Measured with Tektronix SD-24 in 11801 mainframe 20 GHz sampling scope.
Full pulse 1ns/div



Picture 2 - Measured with Tektronix SD-24 in 11801 mainframe 20 GHz sampling scope.
Risetime 200ps/div



Picture 3 - Measured with Tektronix SD-24 in 11801 mainframe 20 GHz sampling scope.
Risetime 100ps/div
Measured risetime is 82ps  (light travels 25 mm in 82ps, a signal on the board about 17mm) 



Picture 4 - Measured with Tektronix 11A71 plugin in DSA602A. (1GHz BW)
risetime 200ps/div



Picture 5 - Measured with Tektronix TDS754A 500MHz scope.
risetime 500ps/div

Measured risetime is consistent with measured bandwidth = 340 / .568 = 619 MHz.




Construction



The important area:




I have attached a pdf file of the schematic

Jay_Diddy_B

[krenzo]

Jay, you can use this program to get screenshots from your 11801.  I use it with my 11801B and a serial-to-usb cable.  I have the source code if you're interested.

[Jay_Diddy_B]

Krenzo,

Interesting!!

The screenshots from the DSA602A are generated by a program that I wrote to convert the Tektronix bitmap file to TIF files. My DSA602A does not have the latest firmware in it.

I was planning on doing the transfer electronically, but for now I use I use a floppy. The 11801 does not have a floppy, so I used a camera. I was going to use GPIB with a prologix adapter. It is not very often that I use the 11801. It is not a very good general purpose scope. 

I will look on the back and see if it has RS232.

Many thanks !!!

Jay_Diddy_B