Thanks so much Patrick and Dennis for this very useful information.
I'm a newbie, and I learn a lot of interesting things on this forum
For insulation, if I can do without thermal paste I prefer.
I will try the Mouser ones you recommended.
Keratherms are expensive I find, 19€ for 10 pieces on store. It takes 16 in all.
I'm thinking of going as simple as possible, and making the version without cascoding.
If I assemble Q1 and Q2 jfets with colson as in the photo on the first page, can I do without TO-92 heatsinks ?
Thierry
I'm a newbie, and I learn a lot of interesting things on this forum
For insulation, if I can do without thermal paste I prefer.
I will try the Mouser ones you recommended.
Keratherms are expensive I find, 19€ for 10 pieces on store. It takes 16 in all.
I'm thinking of going as simple as possible, and making the version without cascoding.
If I assemble Q1 and Q2 jfets with colson as in the photo on the first page, can I do without TO-92 heatsinks ?
Thierry
I am using alumina isolators underneath the MOSFET's. 1 mm thick, using thermal paste on both sides. Did some calculation between alumina of 1 mm thick and standard mica. The alumina is much lower in thermal resistance, even though it is much thicker. With the Keratherm, I had a few bad experiences: a small debris after machining the heat sink penetrated the keratherm and made my mosfet burn. I recently ordered (and received) some Aluminium Nitride isolators in TO-247 size from China (Ali-Express) half a mm thick. Should be even much better than the alumina: thermal conductance of aluminium nitride is very high, much higher than alumina. Did not have a chance to try those. With the alumina, I measured the temperature of the mosfets and the heatsink with a thermocouple. Only a few degrees C, at a bias of 1.2 A on a V2 (0.6 A per mosfet), both temperatures are around 40 C.
The alumina one I use is 1 mm thick, aluminium oxide (like the Aavid). Bought it from Amazon in the Netherlands, TO-247 format. €25 for 100 pieces. Don't know if they also sell smaller quantities. But for me, a person who sometimes leaves small debris and imperfections when drilling or tapping in heat sinks, it works much better than Keratherm (which sometimes gets penetrated by debris) or mica (which has much lower thermal conductivity). Only disadvantage: have to use thermal paste.
Alumina (aluminium oxide, Al2O3) is a white ceramic. Electrical resistivity is 10^14 Ohm.cm. A TO-247 sized piece of 1 mm thickness has an electrical resistance of 5000 GOhm. A fingerprint conducts a thousand times better. Comparison: mica and plastic are around 10^16 Ohm.cm. True, 100 times higher than alumina, but irrelevant. Thermal conductivity of alumina is around 100 times higher than mica.
Silver mica is not used in PCs. I think that's a kind of capacitor, actually. There is no mica used in any thermal compound. You may be thinking of Arctic Silver, which is a specific product from a specific brand that clearly lists that the product is 100% conductive and should not be used on any component where electrical insulation is needed.
Almost all thermal compounds used in PC components (GPUs, for example) use aluminum oxide granules mixed into a base, the granule shape, density and particle characteristics give the compound its various properties. Aluminum Oxide is not aluminum. The oxidation process itself causes complete loss of conductivity.
https://www.aluminummanufacturers.org/aluminum-sulfate/aluminum-oxide/
Maybe this link helps understand the material better. It's not a mixture, or an alloy that can separate into its constituent parts. It's a compound all by itself and can be created by oxidising aluminum (sometimes known as anodising). I do seem to remember Nelson mentioning anodised aluminum as his only insulating layer.
While I've done this myself on occasion, I don't make a habit of it as our oxide layers tend too be 5 to 10um only, which doesn't stand up to repeated usage. We don't have the 20um possible in first world countries. Most cities have banned anodising plants due to the chemicals used and it becomes a logistics issue as well. In any case, Alumina is a perfectly acceptable material for insulators. It is used on the back of ceramic hybrid amplifiers, such as the venerable STK series.
If you download the Aavid datasheet from this page, and find your way to page 104, you'll find the data for their insulators. Aavid wouldn't get things this wrong, I can promise you that.
https://www.digikey.com/catalog/en/partgroup/aluminum-oxide-ceramic/51058
These Aavid platelets look very good, Sangram and Starcat! With the 15 W/mK thermal conductivity that Aavid specifies, and an area of 2 cm2 for a TO-247 package, the temperature jump over 1.78 mm thickness is numerically roughly equal to half the device power in Watts. A dissipation of 30 W in the MOSFET then gives 15 degrees temperature jump over the 1.78 mm thickness. Not bad. Mica has a thermal conductivity of 0.3 W/mK perpendicular to the plane, a factor 50 worse. Even if the mica plate is 0.1 mm thick, the temperature jump will still be twice higher than with 1.78 mm of alumina.
Just finished my F5T V2. 32 V rail, cascoded. First biased at 360 mV over the source resistors, but found relatively high distortion with an 8 Ohm load, starting at 5 W per channel (measured with the FFT option in the oscilloscope). I suspect that the diodes start kicking in: I have them mounted above the pcb (pcb mounted upside down), so they are heated by the heat generated by the Mosfets and rising in the heatsink. The Onsemi datasheet tells me that of the 700 mA bias per device, already 100 mA could flow through the diodes if at 45 C: perhaps that causes the distortion. Reduced the bias to 300 mV, and gone is the distortion. Will do proper distortion measurements tomorrow.
This is the distortion spectrum measured on my fresh F5T. Second harmonic is at -60 dB, third harmonic at -85 dB. Hum (50 Hz and its harmonics) are at an inaudible -80 dB, and does not scale with output power. Measured at 1 W output. This spectrum persists for higher powers, and only near 50 W output power the third harmonic increases.
Source is high purity 1 kHz, Akitika (DIY source of $79), and amp output is measured by a 24 bit sound card. I used a Focusrite 6i6 because I had that lying around, but one can also use a Behringer sound card of $30, connected to a PC with USB. Plus a resistor based attenuator so the input of the sound card is not blown up. The software I used is REW, a free software package designed to measure room acoustics, but is has a RTA facility which generates these graphs, including the automatic measurement of all harmonics, THD+N, etc in the upper left corner. Works like a sun, generates nice measurements without expensive distortion analyzers. Requires some soldering, but everyone on this forum can build these components. My F5T is 32 V rails, 1 kVA toroidal transformer, 5U chassis. Need to fiddle around the internal wiring of the 50 Hz power lines, the present loudness of the hum is not audible, but it should be easy tp improve things with extra twisting of cables.
Source is high purity 1 kHz, Akitika (DIY source of $79), and amp output is measured by a 24 bit sound card. I used a Focusrite 6i6 because I had that lying around, but one can also use a Behringer sound card of $30, connected to a PC with USB. Plus a resistor based attenuator so the input of the sound card is not blown up. The software I used is REW, a free software package designed to measure room acoustics, but is has a RTA facility which generates these graphs, including the automatic measurement of all harmonics, THD+N, etc in the upper left corner. Works like a sun, generates nice measurements without expensive distortion analyzers. Requires some soldering, but everyone on this forum can build these components. My F5T is 32 V rails, 1 kVA toroidal transformer, 5U chassis. Need to fiddle around the internal wiring of the 50 Hz power lines, the present loudness of the hum is not audible, but it should be easy tp improve things with extra twisting of cables.
Newbie here, I wanted to build an F5 V3 Turbo - Monos, so I bought a pair of f5 Turbo V3 boards, a pair of PSU boards, JFET (2 nos) (NP Quad (NNPP))
, and also an f5 Parts kit. Then I started to read the thread. This seems like a lot of variables here. can anyone help me with the BOM for the PSU and F5 parts list please, going to order from either mouser or Digikey. Any pointers and advice as well please.
, and also an f5 Parts kit. Then I started to read the thread. This seems like a lot of variables here. can anyone help me with the BOM for the PSU and F5 parts list please, going to order from either mouser or Digikey. Any pointers and advice as well please.
How experienced are you with amp building? Newbie only in the forum, or in DIY electronics in general?
This is not a beginners amplifier, to quote the store.
First start here... Read it all.
https://firstwatt.com/pdf/art_f5_turbo.pdf
For any responses to be meaningful, you will need to write back with exactly which version you wish to build and provide more information. You are correct that there are a lot of variables; they are yours to choose. Otherwise, it will be challenging to help you. As Andy said, this is not a beginner's amp, particularly the V3. Unless you really, truly, need all that power, I'd strongly suggest you choose another version or perhaps even another amplifier.
At a minimum you should include:
- What rails you wish to run
- What PSU you wish to use
- Total number of output pairs
- What chassis are you using for each monoblock.
- Iq => Dissipation per channel you plan to run?
- What specific JFETs did you purchase and do you know their spec'ed breakdown voltage. How far are you willing to push them?
- Describe the function of diodes (D1-D8) in the schematic on pg. 13. Do you know why a number of builds have released the magic smoke due to their inclusion (coupled with bias settings perhaps a bit close to the edge for some Fearless Amplifier Builders). Are you willing to take this chance? How will you know if your amplifier is stable? Do you know why Nelson jokes (but only a bit) that you may want a fire extinguisher nearby?
If you are at the level you describe, please post your proposed BoM with any questions. It should be no trouble for a person with the ability to build this amp w/o causing themselves problems. You should not need one provided. People will certainly help, but you should be able to come up with the fundamentals.
My pointer / advice echoing Andy ... Once again with feeling, if I could not provide the information requested above, and create a basic BoM; I would not build this particular amplifier / version.
My
Good luck, and have fun!
Hello sanjay13leo,
I fully agree with the statements of andynor and ItsAllInMyHead. This is not a real beginners amp.
Allthough I have built my F5T-Monoblocks as the first poweramps out of the portfolio offered to
us 'greedy boyz' by Nelson Pass.
But I was not an absolute beginner in electronics and mechanical engineering. And I was reading
through approximately 500 pages -in different threads about the F5T and its 'relatives' (F5T V1 /V2 /V3)
- and I made a lot of notes.
I was sure I wanted to run my rails at around +- 40 to 45 Volts DC
- so I have chosen 2 transformers with primaries at 2x 115 V AC (= 230 V AC in connected in series)
and secondaries at 2 x 30 V AC out of the transformers. Each transformer is 1000 W - too big - not necessary
- then I have chosen the diyAudio PSU-boards - one set for each Monoblock. I have built the discrete
Diode-bridges (MUR 3020WT) and caps are 10.000 µF / 80 V.
- I built the symmetrical PSU completely and measured my voltage output after the PSU - I landed anywhere
around +- 42.8 V DC
I knew from the F5T - article from Nelson Pass, that I have to cascode the input JFets on the driver board.
Running railvoltages of over +- 32 V DC need some protection for the JFets.
And I decided that I wanted to drive two N-channel-boards and two P-channel-boards per Monoblock.
So, I decided to use 4 JFets (matched NNPP from the store) + cascode-BJTs per driverboard.
I have bought a matched octet N-P-JFets from the diyAudiostore.
And you should get matched N- and P- channel Mosfets for the outputstage.
And a lot of heatsink....
So you see - many decisions before starting your build
Only some thoughts and repeating andynor and ItsAllInMyHead...
Cheers
Dirk
I fully agree with the statements of andynor and ItsAllInMyHead. This is not a real beginners amp.
Allthough I have built my F5T-Monoblocks as the first poweramps out of the portfolio offered to
us 'greedy boyz' by Nelson Pass.
But I was not an absolute beginner in electronics and mechanical engineering. And I was reading
through approximately 500 pages -in different threads about the F5T and its 'relatives' (F5T V1 /V2 /V3)
- and I made a lot of notes.
I was sure I wanted to run my rails at around +- 40 to 45 Volts DC
- so I have chosen 2 transformers with primaries at 2x 115 V AC (= 230 V AC in connected in series)
and secondaries at 2 x 30 V AC out of the transformers. Each transformer is 1000 W - too big - not necessary
- then I have chosen the diyAudio PSU-boards - one set for each Monoblock. I have built the discrete
Diode-bridges (MUR 3020WT) and caps are 10.000 µF / 80 V.
- I built the symmetrical PSU completely and measured my voltage output after the PSU - I landed anywhere
around +- 42.8 V DC
I knew from the F5T - article from Nelson Pass, that I have to cascode the input JFets on the driver board.
Running railvoltages of over +- 32 V DC need some protection for the JFets.
And I decided that I wanted to drive two N-channel-boards and two P-channel-boards per Monoblock.
So, I decided to use 4 JFets (matched NNPP from the store) + cascode-BJTs per driverboard.
I have bought a matched octet N-P-JFets from the diyAudiostore.
And you should get matched N- and P- channel Mosfets for the outputstage.
And a lot of heatsink....
So you see - many decisions before starting your build
Only some thoughts and repeating andynor and ItsAllInMyHead...
Cheers
Dirk
sanjay13leo,
You've received some very detailed replies and I recommend you go through all the steps in the above post by ItsAllInMyHead.
I'd like to follow up on a point he raised: Do you need this much power? Can you perhaps describe your speakers and size of you listening space?
When one's coming from, say a (large-ish) class A/B amp, one might feel the need to go with a class A amp with a similar power rating.
But the survey here: (https://www.diyaudio.com/community/...h-voltage-power-do-your-speakers-need.204857/ )
shows that many users don't actually use too much amplifier power. So I suggest you first try the test outlined there to give an idea of your power need,
bearing in mind that in general, a large class A amplifier is a non-trivial endeavour.
And please read Nelson's article carefully with especial attention to p.10 and the top half of p.11.
Cheers,
Dennis
You've received some very detailed replies and I recommend you go through all the steps in the above post by ItsAllInMyHead.
I'd like to follow up on a point he raised: Do you need this much power? Can you perhaps describe your speakers and size of you listening space?
When one's coming from, say a (large-ish) class A/B amp, one might feel the need to go with a class A amp with a similar power rating.
But the survey here: (https://www.diyaudio.com/community/...h-voltage-power-do-your-speakers-need.204857/ )
shows that many users don't actually use too much amplifier power. So I suggest you first try the test outlined there to give an idea of your power need,
bearing in mind that in general, a large class A amplifier is a non-trivial endeavour.
And please read Nelson's article carefully with especial attention to p.10 and the top half of p.11.
Cheers,
Dennis