RF amp

On Tue, 23 Jul 2019 00:10:18 GMT, Steve Wilson <no@spam.com> wrote:

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

On Mon, 22 Jul 2019 19:18:32 GMT, Steve Wilson <no@spam.com> wrote:

You are not going to do really fast sstuff on FR4. You need Rogers or
similar, which is very expensive to put on a pcb. So use a small piece of
Rogers for the fast stuff, and a larger copperclad for the supporting
circuits.

I do picosecond stuff on FR4, both dremel'd and then as multilayer
PCBs.

What's the risetime and how far do the signals go.
Click to expand...

Ballpark 50 ps. Of course, the trick is to keep the traces short, no
matter what the dielectric. 1" of 50 ohm microstrip on FR4 is OK
around 50 ps, but I would rather keep a fast run shorter.

You tried mixing Rogers or similar and FR4.
Click to expand...

Well, one of my former engineers did that, and it didn't work anyhow.
My version was plain (flat!) FR4 and it did work.

https://www.dropbox.com/s/v9byymfy6ijh64g/T400_wiggles.JPG?raw=1


You got bowls. You don't have
>that problem with Manhattan and daugterboards.

I wouldn't want to make production boards out of bits and pieces.

If I did have to use a microwave laminate, I probably wouldn't mix it
with FR4, and certainly wouldn't do an asymmetric stackup.

Lately all our boards have a flatness spec. The auto pick-and-place
likes flat boards.

Show us some fast stuff that you've done.

Not yet. Soon.
Click to expand...

--

John Larkin Highland Technology, Inc

lunatic fringe electronics
 
On 23/7/19 2:29 pm, John Larkin wrote:
On Tue, 23 Jul 2019 00:10:18 GMT, Steve Wilson <no@spam.com> wrote:

John Larkin <jjlarkin@highland_snip_technology.com> wrote:

On Mon, 22 Jul 2019 19:18:32 GMT, Steve Wilson <no@spam.com> wrote:

You are not going to do really fast sstuff on FR4. You need Rogers or
similar, which is very expensive to put on a pcb. So use a small piece of
Rogers for the fast stuff, and a larger copperclad for the supporting
circuits.

I do picosecond stuff on FR4, both dremel'd and then as multilayer
PCBs.

What's the risetime and how far do the signals go.

Ballpark 50 ps. Of course, the trick is to keep the traces short, no
matter what the dielectric. 1" of 50 ohm microstrip on FR4 is OK
around 50 ps, but I would rather keep a fast run shorter.
Click to expand...

To be fair, your stuff probably survives loss or impedance discontinuity
better than most continuous RF applications would.
 
John Larkin <jjlarkin@highlandtechnology.com> wrote:

On Tue, 23 Jul 2019 00:10:18 GMT, Steve Wilson <no@spam.com> wrote:

What's the risetime and how far do the signals go.

Ballpark 50 ps. Of course, the trick is to keep the traces short, no
matter what the dielectric. 1" of 50 ohm microstrip on FR4 is OK
around 50 ps, but I would rather keep a fast run shorter.
Click to expand...

What ic family or semiconductor technology do you use to get there?
 
John Larkin <jjlarkin@highlandtechnology.com> wrote:

> I wouldn't want to make production boards out of bits and pieces.

Manhattan is not for production. It is for one-off development, or testing
different versions of the same function such as various ic's, pll's, vco's,
etc.

Once you are satisfied with the results, then you put it on FR4.

Doing it this way is a lot cheaper and faster than starting out on FR4.
 
On 7/22/2019 8:39 AM, Steve Wilson wrote:
Lasse Langwadt Christensen <langwadt@fonz.dk> wrote:

mandag den 22. juli 2019 kl. 09.27.07 UTC+2 skrev Clifford Heath:
On 22/7/19 1:13 pm, John Larkin wrote:
On Mon, 22 Jul 2019 11:15:57 +1000, Clifford Heath
no.spam@please.net> wrote:

On 21/7/19 9:56 am, Piotr Wyderski wrote:
John Larkin wrote:
I get to teach a young engineer my Dremeling secrets. I am not
able to produce decent PCBs that way, but ...

I grew up using a dental drill we had at home, so I'm ok freehand,
but...

I've set a Dremel drill press up with the minimum bit height just at
the bottom of the copper. It's trivial to slide the PCB under it in
nice straight lines, but it really needs two hands. I'm considering
attaching
a foot pedal to the lever that drops the bit down, to free the
second
hand. It would be trivial to achieve results better than John's
freehand, that way.

Clifford Heath.

The copper carving is artistic but fundamentally insignificant. The
"results" that matter are the circuit performance.

Yes, and really straight cuts (without needing a lot of practise) would
help with that.

make a simple pantograph, stylus sets the cutting depth and it would be
easy to follow a straight edge to do lines

Do not chop the copper, especially for high speed circuits.

Use Manhattan style. Glue the ic's to the copperclad with superglue. Bend the
legs up live bug style and solder directly to them. Don't use dead bug style.
It conceals the ic part number and makes it much harder to understand six
months later.

Use a small hole punch to cut circles from copperclad for the standoffs. You
can get a Neiko 02612A Multi-Purpose Power Hole Punch Kit 7 Sizes from 3/32"
to 9/32" from Amazon at

https://www.amazon.com/Neiko%C2%AE-02612A-Multi-Purpose-Power-
Punch/dp/B0002T87CW

It may not cut 0.062 copperclad without great effort. You can get 0.010
copperclad from a number of places - pcb houses, used pc component supply
houses, etc. I got some 12" x 24", 0.014" thick, 1 oz. copperclad from

http://www.surplussales.com/RF/RFMicrowaveCir.html

You can save time troubleshooting crosstalk and ground bounce problems by
placing a noisy circuit on a small piece of copperclad and using standoffs to
isolate the ground noise from the rest of the circuitry. Ground the noisy
circuit at the point that gives the least crosstalk. PLL phase detectorS,
VCO'S, fast comparators, etc benefit from this approach.

You can also use this trick in reverse. Put a sensitive circuit on a small
piece of copperclad. Ground it to the main circuit at the point that gives
the least crosstalk.

Use P. Hobbs trick of coating the copperclad with clear acrylic spray before
starting construction. You can solder through the acrylic with no problems.

You will end up with circuits that are essentially unlimited in size, are
strong enough to be put in an enclosure, have a clean ground plane to
minimize noise, can be easily modfied when needed, can often eliminate the
need for a costly pcb, will look nice, and will last forever.
Click to expand...

As a substitute for the acrylic spray, try MG Chemicals Liquid Tin to
keep the copper from oxidizing. I've used it and it works fine for me.
 
On Tue, 23 Jul 2019 08:30:54 GMT, Steve Wilson <no@spam.com> wrote:

John Larkin <jjlarkin@highlandtechnology.com> wrote:

On Tue, 23 Jul 2019 00:10:18 GMT, Steve Wilson <no@spam.com> wrote:

What's the risetime and how far do the signals go.

Ballpark 50 ps. Of course, the trick is to keep the traces short, no
matter what the dielectric. 1" of 50 ohm microstrip on FR4 is OK
around 50 ps, but I would rather keep a fast run shorter.

What ic family or semiconductor technology do you use to get there?
Click to expand...

There are a few arguably affordable parts that make fast edges.
ADCMP582, and the Onsemi NBxxx (gigacomm) exotica. Hittite (now ADI)
has some gates that get down into the 10 ps range for more money.
There are also some (Russian?) gates in that range; they have a US rep
with a funny name that I can't recall.

Step-recovery diodes can make volts of step with very fast edges. And
there are shock lines, which I haven't tried.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics
 
John Larkin <jjlarkin@highlandtechnology.com> wrote:

Step-recovery diodes can make volts of step with very fast edges. And
there are shock lines, which I haven't tried.

Do you have any web pointer to those "shock lines"?
Click to expand...

Thanks

--
Uwe Bonnes bon@elektron.ikp.physik.tu-darmstadt.de

Institut fuer Kernphysik Schlossgartenstrasse 9 64289 Darmstadt
--------- Tel. 06151 1623569 ------- Fax. 06151 1623305 ---------
 
John Larkin <jjlarkin@highlandtechnology.com> wrote:

On Tue, 23 Jul 2019 08:30:54 GMT, Steve Wilson <no@spam.com> wrote:

What ic family or semiconductor technology do you use to get there?

There are a few arguably affordable parts that make fast edges.
ADCMP582, and the Onsemi NBxxx (gigacomm) exotica. Hittite (now ADI)
has some gates that get down into the 10 ps range for more money.
Click to expand...

What products have you used these devices in?
 
On 23 Jul 2019 14:24:12 GMT, Uwe Bonnes
<bon@hertz.ikp.physik.tu-darmstadt.de> wrote:

John Larkin <jjlarkin@highlandtechnology.com> wrote:

Step-recovery diodes can make volts of step with very fast edges. And
there are shock lines, which I haven't tried.

Do you have any web pointer to those "shock lines"?

Thanks
Click to expand...

google for NLTL, or nonlinear transmission line, or RF comb generator.

I have a sample NLTL comb generator in a little SMA tube, but I
haven't tried it yet.


--

John Larkin Highland Technology, Inc

lunatic fringe electronics
 
On Thursday, July 18, 2019 at 3:39:43 PM UTC-4, John Larkin wrote:
We need 28 volts p-p at 14 MHz to test our IQ modulator box. I've
bought two Chinese RF amps from Amazon and both are garbage. So we're
going to build our own class C amp, as a dremel'd prototype.

https://www.dropbox.com/s/tq8qsvc7lbfirmh/Z468_WB_1.png?raw=1
Click to expand...

Here is a kit 45W HF amplifier that is Push Pull for lower distortion. It is designed for CW, FM, or SSB applications. It does not come with a heat sink, but it"s under $15. Just reduce the drive to get the output that you need, and mount it in a metal box. It is typical of the suggested designs in the old Motorola RF Databooks.

45W-SSB-Linear-Power-Amplifier-CW-FM-HF-Transceiver-Shortwave-DIY-Kit-US

https://www.ebay.com/itm/253977892800
 
You must log in or register to reply here.

Welcome to EDABoard.com

Sponsor

Back
Top