EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: axitece on October 03, 2017, 12:58:23 pm
-
Hello everyone,
I've started working on a DIY oscilloscope analog front end project to develop my analog design skills. I want the front-end to have a BW of ~100 MHz so that it can be built using discrete components.
I've come across a ton of really helpful information on this forum regarding the overall architecture of the signal path. However, I haven't found any detailed information on the design of the attenuators. It seems like most of the information available online is for ladder network based compensated attenuator shown in the "Electronic Instrumentation" (http://books.google.es/books?id=ECbjfoJQ6MoC&pg=PA209&lpg=PA209&dq=compensated+switchable+attenuator&source=bl&ots=LuFRucEQF-&sig=04agda47i8OZk879as1h12LTgjQ&hl=es&sa=X&ei=79vzU6iPEInnygOYwYLgDQ&ved=0CGoQ6AEwCA#v=twopage&q&f=false) book. A similar design is used in the Analog Discovery 2 as shown (https://reference.digilentinc.com/reference/instrumentation/analog-discovery-2/reference-manual#scope_input_divider_and_gain_selection) in its technical reference manual.
Despite the ladder based compensated attenuator being useful and easy to understand, it seems that going with discrete attenuation stages that can be switched in or out of the signal path using DPDT switch is a better idea. In the ladder network based attenuator, the input capacitance will change with attenuation setting as mentioned here (https://www.eevblog.com/forum/projects/advices-with-dso-input-stage-design/msg503643/#msg503643) and here (https://www.eevblog.com/forum/projects/advices-with-dso-input-stage-design/msg503964/#msg503964). This advice is also backed by the fact that most oscilloscopes use discrete attenuators that are switched in or out of the signal path to obtain different attenuation levels. I'd really like to understand the purpose of different resistors and capacitors that are used in these discrete attenuators such as the ones in the Rigol front-ends for DS1052E (http://rigol.codenaschen.de/images/0/0c/DS1052E_HW58_PCB_Schematics_-_Ch1_analog_front-end.jpg) and/or DS1054Z (http://www.eevblog.com/files/Rigol-DS1054Z-Schematic-FrontEnd.pdf). If anyone has a good understanding on the design of these specific attenuators, I'd really appreicate if they can respond since it's a lot better to understand an existing design thoroughly before starting my own from scratch.
It'd be great if someone could point me to any resources or give advice on the design decisions to be made in design of such attenuators: what topology (pi, T, ladder) to use and when, how to design compensation using trimming capacitors, etc. I don't expect all of these questions to be answered but I am hopeful that I'll be able to learn more about these aspects of analog design.
-
Tektronix produced a series of books on the design of oscilloscopes in the 60s. I believe they were called Circuit Concepts and the volume on vertical amplifiers has stuff on attenuators. If memory serves, the X10, X100 & X1000 were implemented at the input using megohm range resistors and pF trimmer caps while the X1, X2 & X5 were implement after the 1st stage using much smaller resistors.
A big problem with implementing a 100 MHz 'scope attenuator is inductance and finding small switches to minimize loop area. Tek, and I believe hp went to little spring leaf switches right on the PCB and actuated by camshafts. Quite the mechanical marvel. BTW, an FM band transmitter only requires a few turns of wire and a few pF of capacitance to form a resonant circuit. I couldn't imagine using a miniature rotary switch to actually switch the signal, maintain isolation and not add to the loop area. Small reed switches might be a way to make this work.
Cheers,
-
I remember reading Paul Rako's account of struggling to create a silicon based variable frontend attenuator for his automotive customer.
He rang Jim Williams.
Jim said: "Go find nearest Tek scope, turn vertical amplifier knob slowly an tell me what you observe." Paul paused (the scope was on his bench) and laughed to which Jim replied: "if Tek could not figure out the way to replace relays then maybe you should not waste your time trying"
So Paul redesigned his frontend and everything was cushdy ever after.
Things have changed since then and 100MHz is probably within reach of silicon based attenuators now.
Relays are still used - I have personally used these: https://www.digikey.com/product-detail/en/standex-meder-electronics/CRF05-1A/374-1041-1-ND/698052 (https://www.digikey.com/product-detail/en/standex-meder-electronics/CRF05-1A/374-1041-1-ND/698052) and these https://www.digikey.com/product-detail/en/coto-technology/9814-03-00/306-1156-5-ND/586519 (https://www.digikey.com/product-detail/en/coto-technology/9814-03-00/306-1156-5-ND/586519)
Unfortunately, good stuff is mostly SPST so attenuator will end up being quite expensive.
Leo
-
Certainly true, but check out this article by Steve Roach of Tektronix, on the design of a relay-less front end:
https://archive.org/stream/fe_The_Art_And_Science_Of_Analog_Circuit_Design_PCB/The_Art_And_Science_Of_Analog_Circuit_Design_PCB#page/n81/mode/2up
-
Tektronix produced a series of books on the design of oscilloscopes in the 60s. I believe they were called Circuit Concepts and the volume on vertical amplifiers has stuff on attenuators. If memory serves, the X10, X100 & X1000 were implemented at the input using megohm range resistors and pF trimmer caps while the X1, X2 & X5 were implement after the 1st stage using much smaller resistors.
A big problem with implementing a 100 MHz 'scope attenuator is inductance and finding small switches to minimize loop area. Tek, and I believe hp went to little spring leaf switches right on the PCB and actuated by camshafts. Quite the mechanical marvel. BTW, an FM band transmitter only requires a few turns of wire and a few pF of capacitance to form a resonant circuit. I couldn't imagine using a miniature rotary switch to actually switch the signal, maintain isolation and not add to the loop area. Small reed switches might be a way to make this work.
Cheers,
Thanks for reminding about the Tektronix Circuit Concepts books! I had skimmed through the Vertical Amplifiers chapters earlier but I didn't remember seeing a section on the input attenuators. I'm reading through it again right now and I must say that it does a great job of explaining important concepts in an easy to understand manner. I'll be sure to comment in this thread with my understanding on the content presented in that book once I'm done.
As far as the switching between different attenuation stages is concerned, you're absolutely correct that solid state switches won't cut it for the required bandwidth. I should've made it clearer in the original post, but I don't plan on using solid state switches but small reed relays for the exact reasons you mentioned.
-
I remember reading Paul Rako's account of struggling to create a silicon based variable frontend attenuator for his automotive customer.
He rang Jim Williams.
Jim said: "Go find nearest Tek scope, turn vertical amplifier knob slowly an tell me what you observe." Paul paused (the scope was on his bench) and laughed to which Jim replied: "if Tek could not figure out the way to replace relays then maybe you should not waste your time trying"
So Paul redesigned his frontend and everything was cushdy ever after.
Things have changed since then and 100MHz is probably within reach of silicon based attenuators now.
Relays are still used - I have personally used these: https://www.digikey.com/product-detail/en/standex-meder-electronics/CRF05-1A/374-1041-1-ND/698052 (https://www.digikey.com/product-detail/en/standex-meder-electronics/CRF05-1A/374-1041-1-ND/698052) and these https://www.digikey.com/product-detail/en/coto-technology/9814-03-00/306-1156-5-ND/586519 (https://www.digikey.com/product-detail/en/coto-technology/9814-03-00/306-1156-5-ND/586519)
Unfortunately, good stuff is mostly SPST so attenuator will end up being quite expensive.
Leo
I mostly definitely will be using relays rather than solid state switches for the attenuator. I'll be sure to keep the relays that you've mentioned in mind when picking the final components. What do you think about this Fujitsu relay (http://www.fujitsu.com/downloads/MICRO/fcai/relays/ftr-b3.pdf)? This was listed as the DPDT relay used in the Rigol DS1054Z front-end circuit (http://www.eevblog.com/files/Rigol-DS1054Z-Schematic-FrontEnd.pdf) that was reverse engineered by Dave. I guess i'm unsure as to what specifications one would care about when selecting a relay for this kind of application where bandwidth is important? I'm very interested in your input since it seems like you've dabbled in this kind of stuff before. Thanks!
-
I guess i'm unsure as to what specifications one would care about when selecting a relay for this kind of application where bandwidth is important?
For oscilloscope it is mostly insertion loss at desired frequency and its flatness. High speed relays tend to [also] quote risetime which can be worked back to bandwidth. Isolation is not really a big deal because scopes don't tend to have a high dynamic range or extreme accuracy so a little bit of leakage across open contacts is not the end of the world. It is usually taken care of by shorting undesired leakage paths to ground with SPSTs.
Leo