TIA-PD compensation, by eye

Hard to say. It depends on where the currents go.

I use copperclad ground plane for prototypes regardless of the frequency.

When you have devices that switch in sub-nanosecond speeds, wideband op amps, low level signals to track, anything to do with a PLL, etc, a solid ground plane is essential. I would rather spend my time improving a design than waste it on troubleshooting grounding problems.

I will also give here a secret I have been keeping for decades.

If you put a noisy circuit on its own copperclad, and use the minimum number of ground connections to the main pcb selected at optimum points, you can cut the crosstalk from the noisy circuit very significantly.

This is especially significant in low level circuits, PLL counters, VCOs, phase detectors, and anywhere else you are working with sensitive circuits.

Once you have the circuit working with minimum noise, you can easily transfer the layout to a working PCB.

This sounds similar to the daughterboards used on some oscilloscope front ends, although they seem to be using it to isolate the input signals instead of reducing crosstalk to or from other circuits. It is not clear how the daughterboard could reduce common mode switching noise from the CPU from going into the scope probe ground.

So the idea is to figure out what the currents are you have to deal with, and where they come from and where they go.

Reply to
Steve Wilson
Loading thread data ...

Hmm that sucks about the customer, have they bought a bunch of other stuff from you?

Anyway I remember your switched cascodes. It's sweet. I'm much more pedestrian... walking along at DC. If this ever is a product it will have to live in the same box. And I really like (use) the ten position switch. (Almost five orders of magnitude!) So that's gotta stay somehow. With lots of signals the user can adjust the max BW. (sweep rate) If I have to live with a few pF of feedback, I'll try the opa827 less noise.. (not as much drive current which is too bad, right you'll suggest a line driver.)

Oh Joerg, I wanted to add that this is just a preamp and there's almost always more gain behind it. It also lives in an Al box, electro-static wise.. though lot's of stuff couples in via the big photodiode.

George H.

Reply to
George Herold

You can still use that with John's solution, it's just that you would need 10 TIAs. Since opamps come in quads it may be feasible.

Another trick would be to have a two-layer 10-position rotary switch. One layer selects from 3-4 individual TIAs and the other layer switches in one or two additional resistors as needed. Sort of a hybrid solution.

The old OPA series is quite good. I have used them last year as well. Noise usually gets in via the supply leads. Without a proper ground plane it is next to impossible to muffle that well enough for a harsh RF environment. Harsh can mean simply a cicty like where John's company is, right down the hill from a major radio tower. Or a person with a GSM phone walks by and just typed in a text message.

The solar panel :-)

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

We've used 61,000 so far, most for switching low-level signals, like thermocouples. They seem fine.

Here's a tester board:

formatting link

which is used to test other boards. That has about 270 of those relays.

I've found reeds to be unreliable. And they are expensive. The Fujitsus are DPDT, too, which is handy.

Morse code? Isn't that a variety of smoke signal?

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

No! They are in the semiconductor industry, which seems to have a tradition of ruthlessness, probably inherited from Schockley and Intel.

The aerospace people are much nicer, and don't steal.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

That's good, more than 10 times the number as most clinical trial.

IME such relays work fine no matter how often used. They failed once they reached 15 years or older, sometimes over 20 years. When that happened then usually a lot of them in the same box had erratic contact. Where you had to exercise all of them numerous times and then they'd work fine for the day. A month later, same spiel.

Sort of.

formatting link

I never got past 25 letters per minute, probably because I have no musicla talent which is supposed to be linked to it. I have used morse code a lot.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

I final follow-up (as my pcb is now an inch deep in rosin flux. :^) The series string was no better than parallel. I guess the good news (some may cry 'sour grapes') is that with the switch I don't need any fancy front ends.

George H.

Reply to
George Herold

We use them in products for BIST and occasionally for gain switching, so they don't operate often. I sure hope they don't start dying one of these years.

They switch fast stuff nicely:

formatting link

We are running an experiment right now, using that giant test board. We will test the switch matrix for leakage current, then run a couple of boards through increasingly nasty reflow and wash processes, and see if we have relay problems. I've been insisting on solvent wash, but people want to see if a lead-free reflow profile followed by water wash is safe.

I never got a ham license, partly because I couldn't learn to receive code (and partly because I don't much like to talk to people.) I also have zero, or negative, musical ability; music mostly annoys me.

Mo is a speech therapist. She occasionally gets an old guy who had a stroke or something and can't talk or understand speech, but can still communicate in code.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics
Reply to
John Larkin

By the time major problems might arise you and I could be in a elder care facility, possibly not remembering Ohm's law :-)

That sure is fast. Most notably no bounce at all.

The relay datasheet says "plastic sealed" amd that moisture sensitivity is "not applicable" but I'd get Fujitsu's blessing on that anyhow.

It's a matter of practice. I learned it by listening a whole lot.

Not so for me. I even have MP3 on my road bike and the mountain bike, for boring sections of trail. Bluegrass, African Reggae and so on. What can instill instant goose bumps with riders I am about to pass is when I switch to the theme melody of "The Good, the Bad and the Ugly". Saw that movie again yesterday.

formatting link

That could very well happen to me some day. I haven't done ham radio in over 25 years but code is as ingrained as the ability to ride a bicycle. Not at the old speeds but that would come back.

It's like the NATO alphabet spelling "Alpha, Bravo, Charlie, Delta ..." which is so ingrained from ham and military days that I instantly use it on noisy phone connections. If the guy at the other end is ex-military or a pilot that instantly works and reduces comm errors to about zero.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

Those pics are TDR of settled relays. I think I have some bounce pics somewhere. They do bounce on closure, for less than a millisecond as I recall. They do not bounce or twang as much as reeds.

We used some second-source Omron parts that had bad sealing and got water inside. If a couple of the giant boards are OK, I might let people use water wash on the Fujitsus.

The spell checker wants to turn Omron into Moron.

Clint was great.

I hate conference calls about tech issues. Some of my customers are always requesting those dreadful group call-ins. People show up late, have to leave, don't identify themselves, and are unintelligible. I have to assume that everyone is taking the same notes. Tech stuff should be email.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

Any recommendation for a dual coil 12VDC DPDT relay with a low insertion loss? I need to make a couple replacement attenuator boards for HP 3325A/B function generators. I have several that were supposed to be complete, but the boards were missing.

Reply to
Michael A Terrell

I'd troll digikey, with a long list you can order by number in stock which limits your choices to something popular with others.

George H.

Reply to
George Herold

The latching version of the Fujitsu has a single coil, but some simple circuit could translate the drive.

Since it transmits a clean 100 ps edge, it's a nice 3 GHz switch. Some of that 100 ps is probably loss on my PCB.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

That doesn't reflect reliability, or other issues. Just what Digikey sells. The original relays lasted for over 20 years, but they are NLA. The board was through hole, a SMD board would be about 1/4 the size.

Reply to
Michael A Terrell

What is the insertion loss form DC to say 100 MHz? It has to be flat, since there is no way to compenstate for variations.

Reply to
Michael A Terrell

Probably close enough to zero that it doesn't matter.

--

John Larkin         Highland Technology, Inc 
picosecond timing   precision measurement  

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com
Reply to
John Larkin

Video conferencing works great IME. If I am the initiator I usually start at the assigned time even if some aren't there yet. Same as I did with real meetings. That has a teachable effect because now the late comers have really missed stuff and often must shamefully admit it at some point.

Zoom works best IMO, next up is GoToMeeting. Some others are sub-par though. If needed the whole thing can be recorded.

What I found really nice is the ability to tie in an oscilloscope or other gear as a "participant" or it can be viewed and operated via someone's screen. That has made a lot of business trips unnecessary.

Probably the only entities online conferencing isn't good for are airlines, rental car companies and hotels.

--
Regards, Joerg 

http://www.analogconsultants.com/
Reply to
Joerg

ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.