I heard somewhere there is a pcb material that is as good as Rogers for
microwaves but not as expensive. Anyone know what this is and where to get
it?

--
MRM

On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
<spamme@not.com> wrote:

>I heard somewhere there is a pcb material that is as good as Rogers for
>microwaves but not as expensive. Anyone know what this is and where to get
>it?

Take a look at Isola Tachyon 100G.

Most good PCB houses will quote it. It fabs like FR4. A rep will
probably send you a biggish sample.

I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
bit better than FR4, but copper loss probably dominated both. The
underside roughness matters as much as the dielectric losses.

On Friday, 9 December 2022 at 22:12:26 UTC, John Larkin wrote:
> On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
> <spa...@not.com> wrote:
>
> >I heard somewhere there is a pcb material that is as good as Rogers for
> >microwaves but not as expensive. Anyone know what this is and where to get
> >it?
> Take a look at Isola Tachyon 100G.
>
> Most good PCB houses will quote it. It fabs like FR4. A rep will
> probably send you a biggish sample.
>
> I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
> bit better than FR4, but copper loss probably dominated both. The
> underside roughness matters as much as the dielectric losses.

Depending on your application, FR4 may be good enough. I recently
tested the attenuation vs frequency of a 10cm long microstrip test
track on an FR4 board that was being used for 10Gbit ethernet.
I don't have all the results to hand, but I just checked a screenshot
that I had captured on my phone. At 3GHz the attenuation of a 10cm long
50 ohm track was about 3dB, increasing uniformly with frequency.
This was a narrow track over a closely spaced ground plane. With wider
tracks and a thicker dielectric layer the attenuation would have been
much lower.

Such attenuation levels may be a problem if you are trying to get really
low noise, or don't want to waste any rf power, but for data transmission
they can be useful in attenuating the reflections from slight
termination mismatches.

John

Am 09.12.22 um 23:12 schrieb John Larkin:
> On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
> <spamme@not.com> wrote:
>
>> I heard somewhere there is a pcb material that is as good as Rogers for
>> microwaves but not as expensive. Anyone know what this is and where to get
>> it?
>
> Take a look at Isola Tachyon 100G.
>
> Most good PCB houses will quote it. It fabs like FR4. A rep will
> probably send you a biggish sample.
>
> I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
> bit better than FR4, but copper loss probably dominated both. The
> underside roughness matters as much as the dielectric losses.

<
https://www.flickr.com/photos/137684711@N07/51843174578/in/datetaken/ >

That's what I got TDRing a 12 mil microstrip on JLCPCB's cheap
7628 4 layer process, not even impedance controlled. €1.60 per
100 * 100 mm board. There cannot be much attenuation, or the
trace would hang to the right. The worst feature of the board
is the SMA connector. I have made a new decal that cuts the ground below
the SMA center pad on layer 2 & 3 and fits a
precision Rosenberger SMA connector. Much, much better.

The JLCpcb impedance calulator says that 50 Ohms is 11.55 mil width.

Gerhard

On Sat, 10 Dec 2022 20:36:32 +0100, Gerhard Hoffmann <dk4xp@arcor.de>
wrote:

>Am 09.12.22 um 23:12 schrieb John Larkin:
>> On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
>> <spamme@not.com> wrote:
>>
>>> I heard somewhere there is a pcb material that is as good as Rogers for
>>> microwaves but not as expensive. Anyone know what this is and where to get
>>> it?
>>
>> Take a look at Isola Tachyon 100G.
>>
>> Most good PCB houses will quote it. It fabs like FR4. A rep will
>> probably send you a biggish sample.
>>
>> I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
>> bit better than FR4, but copper loss probably dominated both. The
>> underside roughness matters as much as the dielectric losses.
>
>
><
>https://www.flickr.com/photos/137684711@N07/51843174578/in/datetaken/ >
>
>
>That's what I got TDRing a 12 mil microstrip on JLCPCB's cheap
>7628 4 layer process, not even impedance controlled. €1.60 per
>100 * 100 mm board. There cannot be much attenuation, or the
>trace would hang to the right. The worst feature of the board
>is the SMA connector. I have made a new decal that cuts the ground below
>the SMA center pad on layer 2 & 3 and fits a
>precision Rosenberger SMA connector. Much, much better.

Yes, connectors usually need plane cutouts. There are some expensive
ones with very skinny pins that match a narrow trace so don't need
plane cuts.

>
>The JLCpcb impedance calulator says that 50 Ohms is 11.55 mil width.
>
>Gerhard

We use the Saturn software to calculate trace impedances and ATLC to
simulate geometries.

https://www.dropbox.com/s/a2weq43lmo1hdch/DSC01527.JPG?raw=1

https://www.dropbox.com/s/hebu7yfzj9mf3lm/Rob1.jpg?raw=1

https://www.dropbox.com/s/1zb71vy9g576c6y/E-field.jpg?raw=1

For digital stuff with short (say, 1" or so) traces FR4 is fine.

I don't think we have any production boards that are anything but FR4,
and we do picosecond stuff.

https://www.dropbox.com/s/jpr0oolz7ewg1u7/T502B6-316.jpg?raw=1

U14 is a 40 GHz switch and U17 is a $300 distributed amplifier.

All on FR4.

On Saturday, 10 December 2022 at 20:13:50 UTC, John Larkin wrote:
> The worst feature of the board
> >is the SMA connector. I have made a new decal that cuts the ground below
> >the SMA center pad on layer 2 & 3 and fits a
> >precision Rosenberger SMA connector. Much, much better.
> Yes, connectors usually need plane cutouts. There are some expensive
> ones with very skinny pins that match a narrow trace so don't need
> plane cuts.

I was once well paid for reducing the pad diameter on a through-the-board
FAKRA connector (an automotive design based on SMB). Reducing the
diameter of the pads on every plane to the minimum that the design rules
allowed gained 0.5dB output from a transmitter.
The customer was happy!

John

John Larkin <jlarkin@highland_atwork_technology.com> wrote:

> On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
> <spamme@not.com> wrote:
>
>>I heard somewhere there is a pcb material that is as good as Rogers for
>>microwaves but not as expensive. Anyone know what this is and where to get
>>it?
>
> Take a look at Isola Tachyon 100G.
>
> Most good PCB houses will quote it. It fabs like FR4. A rep will
> probably send you a biggish sample.
>
> I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
> bit better than FR4, but copper loss probably dominated both. The
> underside roughness matters as much as the dielectric losses.

The roughness also helps the copper adhere to the substrate. This is why it
is so difficult to get smooth, low loss copper to adhere to the substrate.

--
MRM

On Sun, 11 Dec 2022 10:07:55 -0000 (UTC), Mike Monett VE3BTI
<spamme@not.com> wrote:

>John Larkin <jlarkin@highland_atwork_technology.com> wrote:
>
>> On Fri, 9 Dec 2022 21:53:56 -0000 (UTC), Mike Monett VE3BTI
>> <spamme@not.com> wrote:
>>
>>>I heard somewhere there is a pcb material that is as good as Rogers for
>>>microwaves but not as expensive. Anyone know what this is and where to get
>>>it?
>>
>> Take a look at Isola Tachyon 100G.
>>
>> Most good PCB houses will quote it. It fabs like FR4. A rep will
>> probably send you a biggish sample.
>>
>> I TDR tested a microstrip sample, about 20 GHz equivalent. It was a
>> bit better than FR4, but copper loss probably dominated both. The
>> underside roughness matters as much as the dielectric losses.
>
>The roughness also helps the copper adhere to the substrate. This is why it
>is so difficult to get smooth, low loss copper to adhere to the substrate.

Yup. The real microwave teflon stuff can be peeled with tweezers.

This is standard FR4, a high-speed nightmare:

https://www.dropbox.com/s/m65gd96vs8j5qp8/PCB_foil_peel.jpg?raw=1

The Isola is better but not totally shiny, a decent compromise.