Hey, Phil

Do you think this will work?

https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1

We need to snoop the output signal at > 1 GHz bandwidth, preferably
more like 3.

Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
oscillate. I'd like to go lower and faster but it might oscillate, and
the proto layout is kinda ugly so not definitive.

We need roughly -1 volt swing at the phemt source. L1 and C3 could be
stuffing options.

We could run the phemt at 40 mA or so.

Did I mention that I have to finish the PCB layout by tomorrow
morning?
--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

On 12/1/2021 8:01 PM, John Larkin wrote:
>
> Hey, Phil
>
> Do you think this will work?
>
> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>
> We need to snoop the output signal at > 1 GHz bandwidth, preferably
> more like 3.
>
> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
> oscillate. I'd like to go lower and faster but it might oscillate, and
> the proto layout is kinda ugly so not definitive.
>
> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
> stuffing options.
>
> We could run the phemt at 40 mA or so.
>
> Did I mention that I have to finish the PCB layout by tomorrow
> morning?
>

It looks simple enough that it's possible to do the matrix math with the
S or Y parameters and pencil and paper to find out if the structure is
likely to oscillate at the new operating point at least, if the
manufacturer is kind enough to provide that sort of stuff with some data
points that're appropriate for the OP and bandwidth in question. I
understand that's a big "if", though.

John Larkin wrote:
>
> Hey, Phil
>
> Do you think this will work?
>
> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>
> We need to snoop the output signal at > 1 GHz bandwidth, preferably
> more like 3.
>
> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
> oscillate. I'd like to go lower and faster but it might oscillate, and
> the proto layout is kinda ugly so not definitive.
>
> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
> stuffing options.
>
> We could run the phemt at 40 mA or so.
>
> Did I mention that I have to finish the PCB layout by tomorrow
> morning?
>

I normally don't run them that hot, so I don't know what the stability
properties are up there, but it's really hard to get them to oscillate
at 15 mA or so. (*) The output resistor is probably a bit on the big
side, because 1/g_M is probably at least 10 ohms even up at 40 mA.

The only other thing is the pHEMT's poor linearity as a follower, which
will distort the output some. For following pulses, it's maybe not a
big deal.

Cheers

Phil Hobbs

(*) I've used them to run a bootstrap over a 50-mm FFC cable, believe it
or not. Trashed the bandwidth, but it worked like the bomb for the
customer. It's been in production for a couple of years now.

--
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
http://hobbs-eo.com

On Wed, 1 Dec 2021 20:13:01 -0500, bitrex <user@example.net> wrote:

>On 12/1/2021 8:01 PM, John Larkin wrote:
>>
>> Hey, Phil
>>
>> Do you think this will work?
>>
>> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>>
>> We need to snoop the output signal at > 1 GHz bandwidth, preferably
>> more like 3.
>>
>> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
>> oscillate. I'd like to go lower and faster but it might oscillate, and
>> the proto layout is kinda ugly so not definitive.
>>
>> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
>> stuffing options.
>>
>> We could run the phemt at 40 mA or so.
>>
>> Did I mention that I have to finish the PCB layout by tomorrow
>> morning?
>>
>
>It looks simple enough that it's possible to do the matrix math with the
>S or Y parameters and pencil and paper to find out if the structure is
>likely to oscillate at the new operating point at least, if the
>manufacturer is kind enough to provide that sort of stuff with some data
>points that're appropriate for the OP and bandwidth in question. I
>understand that's a big "if", though.

I don't do s-parameter math... I'm a time domain guy. And there are
resistors and caps and traces and stuff involved.

I think Phil knows about using the SAV parts as followers.

Looks like I can't find an L1 value that helps, so that's gone.

Here's an xray of the SAV541. Those wirebonds are scary.

https://www.dropbox.com/s/qttp09vpw3kcee3/SAV541_Xray.jpg?raw=1

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

On Wed, 1 Dec 2021 20:24:45 -0500, Phil Hobbs
<pcdhSpamMeSenseless@electrooptical.net> wrote:

>John Larkin wrote:
>>
>> Hey, Phil
>>
>> Do you think this will work?
>>
>> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>>
>> We need to snoop the output signal at > 1 GHz bandwidth, preferably
>> more like 3.
>>
>> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
>> oscillate. I'd like to go lower and faster but it might oscillate, and
>> the proto layout is kinda ugly so not definitive.
>>
>> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
>> stuffing options.
>>
>> We could run the phemt at 40 mA or so.
>>
>> Did I mention that I have to finish the PCB layout by tomorrow
>> morning?
>>
>
>I normally don't run them that hot, so I don't know what the stability
>properties are up there, but it's really hard to get them to oscillate
>at 15 mA or so. (*) The output resistor is probably a bit on the big
>side, because 1/g_M is probably at least 10 ohms even up at 40 mA.
>
>The only other thing is the pHEMT's poor linearity as a follower, which
>will distort the output some. For following pulses, it's maybe not a
>big deal.
>
>Cheers
>
>Phil Hobbs
>
>(*) I've used them to run a bootstrap over a 50-mm FFC cable, believe it
>or not. Trashed the bandwidth, but it worked like the bomb for the
>customer. It's been in production for a couple of years now.

The data sheet says 392 mS typ at 60 mA, which would be 2.5 ohms, so
I'd expect 4ish at maybe 45 mA, R4=100. I could reduce R3 a bit.

I was surprised when Jonathan reported 140 ps edges with R2 = 499. The
source-bootstrapped gate capacitance must be nil.

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

torsdag den 2. december 2021 kl. 02.01.45 UTC+1 skrev John Larkin:
> Hey, Phil
>
> Do you think this will work?
>
> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>
> We need to snoop the output signal at > 1 GHz bandwidth, preferably
> more like 3.
>
> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
> oscillate. I'd like to go lower and faster but it might oscillate, and
> the proto layout is kinda ugly so not definitive.
>
> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
> stuffing options.
>
> We could run the phemt at 40 mA or so.
>
> Did I mention that I have to finish the PCB layout by tomorrow
> morning?

do you need the gain? http://jahonen.kapsi.fi/Electronics/DIY%201k%20probe/

On Wed, 1 Dec 2021 17:56:18 -0800 (PST), Lasse Langwadt Christensen
<langwadt@fonz.dk> wrote:

>torsdag den 2. december 2021 kl. 02.01.45 UTC+1 skrev John Larkin:
>> Hey, Phil
>>
>> Do you think this will work?
>>
>> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>>
>> We need to snoop the output signal at > 1 GHz bandwidth, preferably
>> more like 3.
>>
>> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
>> oscillate. I'd like to go lower and faster but it might oscillate, and
>> the proto layout is kinda ugly so not definitive.
>>
>> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
>> stuffing options.
>>
>> We could run the phemt at 40 mA or so.
>>
>> Did I mention that I have to finish the PCB layout by tomorrow
>> morning?
>
>do you need the gain? http://jahonen.kapsi.fi/Electronics/DIY%201k%20probe/

No, the monitor could be maybe 0.1 of the signal we're snooping.

Passive resistive scope probes, like the HP54006A, can be crazy fast
and very clean. But I don't want to load my signal much, and the
customer has radical requirements for reverse isolation. I need an
active device in there.

https://www.dropbox.com/s/lxq4ujmkvo755uy/HP54006_probe.zip?dl=0

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

On Wed, 01 Dec 2021 17:01:34 -0800, John Larkin
<jlarkin@highland_atwork_technology.com> wrote:

>
>Hey, Phil
>
>Do you think this will work?
>
>https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>
>We need to snoop the output signal at > 1 GHz bandwidth, preferably
>more like 3.
>
>Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
>oscillate. I'd like to go lower and faster but it might oscillate, and
>the proto layout is kinda ugly so not definitive.
>
>We need roughly -1 volt swing at the phemt source. L1 and C3 could be
>stuffing options.
>
>We could run the phemt at 40 mA or so.
>
>Did I mention that I have to finish the PCB layout by tomorrow
>morning?

Deadline moved to Monday.

The SAV-541 really wants to oscillate as a source follower. We think
it's too much risk. We're considering using a MMIC as our pickoff
amplifier. They are remarkably stable.

--

If a man will begin with certainties, he shall end with doubts,
but if he will be content to begin with doubts he shall end in certainties.
Francis Bacon

John Larkin wrote:
> On Wed, 01 Dec 2021 17:01:34 -0800, John Larkin
> <jlarkin@highland_atwork_technology.com> wrote:
>
>>
>> Hey, Phil
>>
>> Do you think this will work?
>>
>> https://www.dropbox.com/s/kou5507r7y31vmu/Phemt_Follower.jpg?raw=1
>>
>> We need to snoop the output signal at > 1 GHz bandwidth, preferably
>> more like 3.
>>
>> Jonathan hacked a test case with R1=4K and R2=499r, and that doesn't
>> oscillate. I'd like to go lower and faster but it might oscillate, and
>> the proto layout is kinda ugly so not definitive.
>>
>> We need roughly -1 volt swing at the phemt source. L1 and C3 could be
>> stuffing options.
>>
>> We could run the phemt at 40 mA or so.
>>
>> Did I mention that I have to finish the PCB layout by tomorrow
>> morning?
>
> Deadline moved to Monday.
>
> The SAV-541 really wants to oscillate as a source follower. We think
> it's too much risk. We're considering using a MMIC as our pickoff
> amplifier. They are remarkably stable.
>

I don't recall ever running one that hot, but they're remarkably stable
at lower current--15 mA or so.

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
http://hobbs-eo.com