I was discussing my current blues harp preamp design with my dad, and he mentioned that Borbely may have had some concern with FR4 boards...maybe a non-linearity of the material's dielectric constant with varying electric field strengths...something I wouldn't think of as being a huge issue in the audio spectrum, but...I wouldn't know.
I'm wondering if anyone is familiar with his comments on the subject, and maybe could furnish a link to an article somewhere?
Sorry if this is in the wrong place.
I'm wondering if anyone is familiar with his comments on the subject, and maybe could furnish a link to an article somewhere?
Sorry if this is in the wrong place.
That's more of a concern for GHz frequencies than for audio, but there are lower loss materials.
What is your circuit like? If tube, maybe it could be built point to point, etc. Air is the best material.
More info: https://resources.altium.com/p/guide-low-dk-pcb-materials
What is your circuit like? If tube, maybe it could be built point to point, etc. Air is the best material.
More info: https://resources.altium.com/p/guide-low-dk-pcb-materials
I assume in this case it is for vibration control. A teflon pcb is noticeably softer/deader than FR4.
Several of his "Top of the Line Kits" were supplied only with teflon pcbs and
for several others it was an optional upgrade.
This excerpt is from the All-JFET MM/MC Phono Preamp article in AudioXpress 7/2005.
"The all-FET phono preamp is a very sensitive amplifier and is capable of picking
up small signals not only through its inputs but also through vibration. It is,
therefore, very important to make the whole amp as "dead" as possible for vibration
by using rubber stand-offs or mounting the whole board on a teflon plate, 6-8mm
thick. And a mumetal box will certainly put the icing on the cake."
This preamp was from the low price Starter Kits Series and did not offer a teflon pcb as an option.
Several of his "Top of the Line Kits" were supplied only with teflon pcbs and
for several others it was an optional upgrade.
This excerpt is from the All-JFET MM/MC Phono Preamp article in AudioXpress 7/2005.
"The all-FET phono preamp is a very sensitive amplifier and is capable of picking
up small signals not only through its inputs but also through vibration. It is,
therefore, very important to make the whole amp as "dead" as possible for vibration
by using rubber stand-offs or mounting the whole board on a teflon plate, 6-8mm
thick. And a mumetal box will certainly put the icing on the cake."
This preamp was from the low price Starter Kits Series and did not offer a teflon pcb as an option.
Here's my take on this. It's based on my experiences in performing low-level measurements in a lab environment. I found that the greatest contributor to noise in high-gain circuits was the cables. Striking or just wiggling them would induce voltages that completely swamped-out the signals I was interested in.
This is a well-known phenomenon. You can buy cables that generate much lower noise (for a premium price) to see if doing the other stuff might be worthwhile. It would be relatively inexpensive compared to converting a PCB from FR4 to teflon or Rogers. Potential noise from cables is the low-hanging fruit, why reach higher if the lower-hanging stuff doesn't ring your bell?
This is a well-known phenomenon. You can buy cables that generate much lower noise (for a premium price) to see if doing the other stuff might be worthwhile. It would be relatively inexpensive compared to converting a PCB from FR4 to teflon or Rogers. Potential noise from cables is the low-hanging fruit, why reach higher if the lower-hanging stuff doesn't ring your bell?
Exotics like Teflon, cause other problems - the copper does not stick well.
I am not aware of FR4 Er changing with voltage, it does fall a little by 1 GHz and above, but not significant for audio.
I am not aware of FR4 Er changing with voltage, it does fall a little by 1 GHz and above, but not significant for audio.
Biggest contributor to DK changes in FR4 is humidity. The stuff absorbs water. For high voltage circuits, conformal coating is often required.
I'd think that folks who have made their ultra-low distortion oscillators and notch filters would have something to say about FR4 if its Er did have a significant voltage coefficient. I don't recall seeing anything to that effect.
FR4's Er vs. frequency dependence is well known among those who design and build high frequency RF circuits. That said, I'm pretty certain that those $100 VNA's, advertised to be usable up to & beyond 3GHz, use FR4. Some (all?) of them even use PCB traces to make the directional couplers!
I have built and used a few high voltage circuits that went on FR4 PCBs. The HV is about 1KV, for photomultiplier tubes. And the voltage divider board, soldered to the PMT pins, sure looks like FR4 to me. But the important thing is the voltage gradient. It's pretty well distributed on the divider board, with about 100V across each 1206-sized SMT resistor. But I DID thoroughly clean the board after re-assembling it......you betcha. I have done the same with the pulse-processor boards.
FR4's Er vs. frequency dependence is well known among those who design and build high frequency RF circuits. That said, I'm pretty certain that those $100 VNA's, advertised to be usable up to & beyond 3GHz, use FR4. Some (all?) of them even use PCB traces to make the directional couplers!
I have built and used a few high voltage circuits that went on FR4 PCBs. The HV is about 1KV, for photomultiplier tubes. And the voltage divider board, soldered to the PMT pins, sure looks like FR4 to me. But the important thing is the voltage gradient. It's pretty well distributed on the divider board, with about 100V across each 1206-sized SMT resistor. But I DID thoroughly clean the board after re-assembling it......you betcha. I have done the same with the pulse-processor boards.
Thanks everyone for your input. Basically, what I'm reading confirms that FR4 will be just fine. As stated, I'm aware that it can be problematic near and past the GHz range (especially in terms of stub filters, etc), and figured it shouldn't be a huge issue in the audio range (about as low as it goes).
To answer a question about the design, it's pretty simple: 3 stage sub-miniature pentode gain stages, and a JFET source-follower after the final that can drive the output transformer a little harder than the tubes. The gain stages are operated outside their linear region, and the whole thing is very much not a hifi device--it's a preamp meant specifically for people who play blues harmonica.
Re: noise, the cables--and more specifically ground loops through the instrument cables will probably be the biggest source. The regulators will be pretty quiet--I chose to use a Salas UltraBiB v1.3 for the B+...probably overkill, but it's what I went with.
Re: vibration (thanks for the excerpt, @DaveG ), that'll certainly be a problem, because my chosen tubes are super microphonic, but I'll be aiming for the low-hanging fruit of big isolation mega-booties for the enclosure, lol. O-Rings around the tubes will help, too.
All in all, we'll see where I end up after taking care of the low-hanging fruit (e.g. just having an enclosure, at this juncture, will help a lot re: RFI from the LED bulbs around the bench).
Anyways, thanks again for the input. Much appreciated.
To answer a question about the design, it's pretty simple: 3 stage sub-miniature pentode gain stages, and a JFET source-follower after the final that can drive the output transformer a little harder than the tubes. The gain stages are operated outside their linear region, and the whole thing is very much not a hifi device--it's a preamp meant specifically for people who play blues harmonica.
Re: noise, the cables--and more specifically ground loops through the instrument cables will probably be the biggest source. The regulators will be pretty quiet--I chose to use a Salas UltraBiB v1.3 for the B+...probably overkill, but it's what I went with.
Re: vibration (thanks for the excerpt, @DaveG ), that'll certainly be a problem, because my chosen tubes are super microphonic, but I'll be aiming for the low-hanging fruit of big isolation mega-booties for the enclosure, lol. O-Rings around the tubes will help, too.
All in all, we'll see where I end up after taking care of the low-hanging fruit (e.g. just having an enclosure, at this juncture, will help a lot re: RFI from the LED bulbs around the bench).
Anyways, thanks again for the input. Much appreciated.
Sorry, I had assumed folks were familiar with different types of PCB substrates. But it (and teflon) are pretty-much aimed at RF applications. Now you know 🙂
JLCPCB offers Rogers but IIRC only with 2-layer boards. It's noticeably more expensive than plain-vanilla FR4.
Rogers materials aren’t something you can make a multilayer board out of. It comes from the factory copper-clad, and getting the copper to stick is their secret sauce. You can build FR4 layers on one side typically, using the Rogers board as a seed. Good option where you have RF and digital on the same board. US board houses make you supply the Rogers material, then build the FR4 layers to spec.
The really big high volume outfits make RF (and very high speed digital) capable laminates out of stable materials with insane numbers of layers, and feature sizes like 20 micron, blind and buried vias and internal heat sinking coins are the norm. The stuff has higher dielectric absorption than Rogers, but more stable properties than FR4. Cell phones are chock full of this stuff. All processed with PCB techniques, on pretty much the same lines. The only real problem is you can’t push a lot of power through it. The smaller you make the conductors, vias, solder balls…… It’s like trying to wire your house up with wire-wrapping wire. For small signal and maybe a half watt - it’s great. NRE for a run like this in the states is OUTRAGEOUS (think hundreds of $K, same as a prototype wafer run) but overseas it may be doable for DIY, if you already have the resources committed to design it.
The really big high volume outfits make RF (and very high speed digital) capable laminates out of stable materials with insane numbers of layers, and feature sizes like 20 micron, blind and buried vias and internal heat sinking coins are the norm. The stuff has higher dielectric absorption than Rogers, but more stable properties than FR4. Cell phones are chock full of this stuff. All processed with PCB techniques, on pretty much the same lines. The only real problem is you can’t push a lot of power through it. The smaller you make the conductors, vias, solder balls…… It’s like trying to wire your house up with wire-wrapping wire. For small signal and maybe a half watt - it’s great. NRE for a run like this in the states is OUTRAGEOUS (think hundreds of $K, same as a prototype wafer run) but overseas it may be doable for DIY, if you already have the resources committed to design it.
At one time we used Nelco materials for a 10GHz fibre channel simulator test card
https://www.raypcb.com/nelco-pcb/
https://www.raypcb.com/nelco-pcb/
Rogers is not (at least for me) a familiar name (grade designation) for a PCB material compositions, it's apparently an R and TM registered used by Rogers Corporation, which restricts its use by others.
https://en.wikipedia.org/wiki/Printed_circuit_board#Materials
https://en.wikipedia.org/wiki/FR-4
Nope:
It looks to me like anyone who wants to spend the money and has the right PCB design can use Rogers.
If you use it in a design I'd be interested to know if it makes a discernable difference for audio. The human auditory system is very forgiving in some ways but very picky in others..
Not all Rogers' products are PTFE (Teflon) based. Their RO and TMM products are based on thermoset resins and process like FR-4.
You can make multilayer boards from PTFE laminates including mixing it with FR-4. It's expensive (although not as bad as it was in 90's) but it definitely is done. Getting things to stick to Teflon is the sticky wicket. That's what's needed for via holes on Teflon. The current process is an oxygen plasma etch. Another process is to use a chemical sometimes called tetra etch.
Unless you're trapping your signal between a trace and a groundplane, the dielectric constant seen by the signal is pretty close to unity since most of the field is in air.
You can make multilayer boards from PTFE laminates including mixing it with FR-4. It's expensive (although not as bad as it was in 90's) but it definitely is done. Getting things to stick to Teflon is the sticky wicket. That's what's needed for via holes on Teflon. The current process is an oxygen plasma etch. Another process is to use a chemical sometimes called tetra etch.
Unless you're trapping your signal between a trace and a groundplane, the dielectric constant seen by the signal is pretty close to unity since most of the field is in air.