negative voltage regulator

An 800MHz opamp on white board? The kind where you poke stripped wires into? Yikes! That's like bicycling on a sheet of ice with regular slicks as tires.

--
Regards, Joerg 

http://www.analogconsultants.com/

P.S.: Most negative regulators have similar architechtures to LDOs and that's usually a recipe for instability. They often need some ESR in the bigger output caps to remain stable.

--
Regards, Joerg 

http://www.analogconsultants.com/

No, no I was just checking the offset on w. proto. The real thing is a mess over copper clad, I'm pretty sure I just need an input cap. I forgot to turn off the low voltage supply and blew the AD8801 with -15V. (shorting the 3-t reg. :^(

George H.

AD8001, oops

800 MHz current mode amp??? White proto??? Oh, The Horror! 8001 is one of the older ADI current-mode parts, a tad rude and crude maybe. It could be oscillating at a zillion MHz, and that would explain the probing offset. Try touching other nodes with a pencil or a small screwdriver or something.

These things need a tight PCB layout and close bypassing. The 1K input resistor might be a bit much. There's no point to matching input resistances to improve DC offset on current-mode amps; the inputs are very asymmetric.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Yeah, a 79L05 might not like a small ceramic output cap. Add a tantalum or aluminum across the ceramics.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Fast CFB amps are not one of those cases where you can blow off the warnings on the data sheet. Imagine what your PSRR looks like at a few hundred MHz, and you can imagine how important *local* bypassing is.

The TI (nee National) data sheets for the similar LMH670x parts (e.g., LMH6702) do a good job at telling you what to expect and what to watch out. They are basically the same parts with the same performance, but I think TI's docs are better in this case.

And yes, it's unlikely you will good results on a breadboard. Throw that away and get some bare copper-clad board for prototyping. If you like, you can gold-plate it the way JL does. :) Even on a real PCB the layout can be twitchy.

-- john, KE5FX

RFI EMI ??

Daniel T.

Does anyone use stripboard or SOIC carrier board islands on a Manhattan prototype? Thank you in advance.

73,

--
Don Kuenz, KB7RPU

I do, and it's good enough for things like EclipsLite and AD8001 type amps.

We actually delivered two of that last one to a customer.

I made a call to a potential new customer last month, proposing a very improbable high-voltage pulser. They would have been skeptical of it ever working, but I brought along one of my little gold proto boards that did work. I hear that they were impressed. People get tired of PowerPoints.

--

John Larkin         Highland Technology, Inc 

lunatic fringe electronics

Tants are IMHO not reliable enough and aluminum are off limits for hi-rel. Problem is, datasheets are often quite silent about this. Maybe because it wouldn't look good.

--
Regards, Joerg 

http://www.analogconsultants.com/

OK, don't laugh.

The AD8001 is the dip in the middle with 1 k ohm strapped across. There is about 1" of bus wire on the input that might be getting some RF pickup. I tired reducing the input R to 500 ohms, did nothing.

I'll try some better bypassing.

George h.

A small resistance in series with the cap can work, but you have to be careful, because an output short can blow up the resistor if you don't design for that.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
https://hobbs-eo.com

That's the problem, if you have a board with lots of bypass caps it isn't enough to provide one extra R-C series to ground, the resistor would have to be in series with the regulator output. That makes the rail mushy.

I try to avoid LDOs like the plague. Only a few are designed correctly with the proper compensation.

--
Regards, Joerg 

http://www.analogconsultants.com/

Which middle? Need a pointer or xy coordinates.

Ground plane +1 Long leads 0 messy external wiring 0

Bypass caps are basically useless for rf. SRF at or below 1MHz is no help at 1GHZ.

It's not apparent from your layout what you are trying to do, but at least you are doing it on a copper groundplane :)

Think of where the currents are trying to go, and make it easier for them to get there.

You must have a very good soldering iron to spread so much solder everywhere. Remember solder has much poorer conductivity than copper.

Recall even a copper plane has very significant impedance at rf. There is no such thing as ground at high frequencies.

Overall improvement +1

You have significantly changed your prototyping from a year ago. Now you are in territory where even the experts will have difficulty to help.

You also need a much better scope to work with fast chips. And better high frequency probing techniques. The AD8001 is a 800MHz chip. They don't mess around with 100MHz scopes.

Listen to the theory and layout recommendations. From the datasheet:

Line lengths on the order of less than 5 mm are recommended.

Interesting problem. Thanks for sharing.

There ARE space qualified tantals. You just have to accept that they are 6 times the volume of normal ones. And the price. No series resistors needed.

cheers, Gerhard

how do they stop them going bang?

NT

I've been making ugly pcb's like this for years now. (Sometimes I may reuse them... so there is some solder build up.)

Here's the fast part of the circuit.

The spad (diode) is in the black tube thing. At the other end is a laser diode (run as an led at low current.) The spad is not hooked up in the pic I posted.

Here's the pulses.

That's with a one second persistence. Some are kinda weird looking, after pulsing or something. (not sure.)

The rise/fall time may be limited by my 40 MHz 'scope. I'll dig out the faster one.

Oh so I bought three photodiodes from Laser Components. This is the

2nd best one, I can see some 'real' photon detection, but the dark count rate is ~10k to 20k per second! Thebest one (spendy) is supposed to be better. (And cooling helps a lot.... I'll be thinking about how to add a TEC this weekend.)

But I want to try blowing this one up before I put the spendy one in. So far it's been fine. I've got an lnd150 with 10 kohm g-s resistor in the HV line and that limits current to ~100uA.

George H.

Forgot one:

Live bug +1

I gave up on dead bug long ago.