Member
Joined 2009
Paid Member
Hi,
I want to build an amplifier with a FET source-follower output, operated single ended and so a familiar sight in this forum. I simply would like some good advice on choice of a suitable FET
I will operate it with low Vds (e.g. 5V) and idle current should be consistent with providing 10W into nominal 8 Ohm loads.
I don't want to parallel multiple devices, it should be a single device. MOSFET (vertical or lateral) and JEFTs, and SiC are all OK solutions.
The output impedance of the gate drive circuit will be relatively high, e.g. 1K so a FET with low gate capacitance is very desirable.
No nfb will be employed (other than it being a source follower).
Thoughts ???
I want to build an amplifier with a FET source-follower output, operated single ended and so a familiar sight in this forum. I simply would like some good advice on choice of a suitable FET
I will operate it with low Vds (e.g. 5V) and idle current should be consistent with providing 10W into nominal 8 Ohm loads.
I don't want to parallel multiple devices, it should be a single device. MOSFET (vertical or lateral) and JEFTs, and SiC are all OK solutions.
The output impedance of the gate drive circuit will be relatively high, e.g. 1K so a FET with low gate capacitance is very desirable.
No nfb will be employed (other than it being a source follower).
Thoughts ???
Last edited:
Hmm, 10W (RMS) into 8 ohms requires a waveform at the load whose RMS voltage is 8.94V and whose peak value is plus-or-minus 12.65 volts. I suggest you allow at least 4 volts for the source follower VGS drop, thus the power supplies are +16.65 and -12.65 if the source follower is N-channel. It is very difficult to see how you can meet your power-into-8-ohm-load spec with Vds <= 5 volts, when the source follower has +16.65 on its drain and -12.65 on its source.
Member
Joined 2009
Paid Member
I will restrict the voltage across drain-source not by having low power supply rails but through the use of active devices such as coscode. I mention the low Vds because I assume it's an important parameter to know in helping recommend a linear FET.
Last edited:
The IRF-510 (DATASHEET AND PURCHASE INFO FROM DIGI KEY) appears to meet your needs
Better yet, Linear Technology ships a SPICE simulation model of it, in their "standard.mos" library with their LTSPICE simulator.
Although I confess I don't understand your Ninja Circuit Trickery
- VDS = 100V
- IDS = 5.6 A continuous
- Qgate = 8.3 nC
- CISS = 180 pF
- PD(max) = 43 watts
Better yet, Linear Technology ships a SPICE simulation model of it, in their "standard.mos" library with their LTSPICE simulator.
Although I confess I don't understand your Ninja Circuit Trickery
I will restrict the voltage across drain-source not by having low power supply rails but through the use of active devices such as coscode
since I don't see how a "coscode" can help a source follower. Thus I have an opportunity to learn!!
Last edited:
If you are prepared to use cascode, and only wants 10W into 8R, then there is no better than LU1014 for linearity (see Zen Version 9).
May not be low enough Zout, so you need to play some tricks, which adds some complications.
But not impossible.
If you want low Zout, then you need to use a Semisouth or UHC Mosfet with high Yfs.
Or use 500x 2SK170BL in parallel (with cascode).
Patrick
May not be low enough Zout, so you need to play some tricks, which adds some complications.
But not impossible.
If you want low Zout, then you need to use a Semisouth or UHC Mosfet with high Yfs.
Or use 500x 2SK170BL in parallel (with cascode).
Patrick
Member
Joined 2009
Paid Member
A low output impedance is not critical, the types of speakers I plan to use don't perform at their best with high damping factors. A single FET source follower will very likely meet my needs in this regard. The SEWA amplifier by MadsK is a good example but I think there are even more examples from Nelson...
Vladimirk, you've travelled this road a lot further than I, so I pay attention to your recommendation. But I don't see where these devices are available, unless you are offering to supply a few ???
Transistormarkj, The IRF-510 specs look nice, but the Vgs vs Id curves look, well, really curved ! - not at all linear ?
Cascoding the device reduces power dissipation allowing use of smaller device with lower capacitance.
Hannes - which lateral FET do you like the look of ?
Vladimirk, you've travelled this road a lot further than I, so I pay attention to your recommendation. But I don't see where these devices are available, unless you are offering to supply a few ???
Transistormarkj, The IRF-510 specs look nice, but the Vgs vs Id curves look, well, really curved ! - not at all linear ?
Cascoding the device reduces power dissipation allowing use of smaller device with lower capacitance.
Hannes - which lateral FET do you like the look of ?
EUVL, I agree, these parts hardly fit demands of NoGNFB. But with carefully arranged not deep current NFB, they will work perfectly. All basic rules for correct implementation of NFB will be fulfilled - intrinsic linearity, enormously high speed of the output stage, 20dB of NFB, one pair of paralleled KP907A will be enough. I have already designed so with paralleled KP903A jFETs output. The KP907A are special linear MOSFETs, and they are much faster.
At the same time, I can not be shure what is better for sound - the jFETs or faster MOSFETs. I do not know, and have not time to test everything.
Another point, KP907A are available, but much more expensive, something from 10 to 20 USD per pcs.
Last edited:
For NoGNFB design I would propose to look at RD100HHF1 MOSFETs. With close to 4S transconductance one gets 0,25 Ohms Zout for the follower. I used them with good results just at source follower output stage.
If one has NFB design where KP907A are needed, I could arrange shipment, but it could be justified for quantity more than 10pcs and will take near one month. They are available from many sources in russia, but in small quantities (usually only few pcs) and they do not ship for export. Only buying via partners are possible.
Last edited:
Member
Joined 2009
Paid Member
Member
Joined 2009
Paid Member
ooops, I misread the IRF510 datasheet - it's a log current scale so I can't really see how linear the thing is. However, the gate capacitance looks to be one of the lowest for a MOSFET, better than the 2SK1058 lateral. The IRF510 also seems to have a lower input capacitance than the RD100H. For a buck each it looks like a good starting point.
Member
Joined 2009
Paid Member
The front end of the amplifier will be a single vacuum tube - all the voltage amplification will come from this tube operating in common cathode and the anode will be directly coupled to the FET gate without a cap or transformer. The anode load very likely be passive, I have chokes good to 8mA of dc.
Hence, the source impedance will be relatively high compared with the majority of SS approaches. A value of around 1k - 3k would be my estimate.
fyi - The cascode device is Bipolar (already have devices riveted to some old heatsinks) and it's base will follow the output signal in order to keep the Vgs of the source follower roughly constant (e.g. at 5V) and at a low value to minimize power dissipation giving me flexibility to use a smaller FET with low input capacitance.
The source follower will have an active source load (again, Bipolar - already have devices riveted to some old heatsinks). The simple approach suggests constant current source, but it will be tempting to play with some other choices such as Aleph / gyrator.
I've never built such an amplifier before and the only building block from those I've just described that I have used before is a common cathode triode, never used a FET before, so part of the fun here is doing something new even if it ain't perfectly engineered
Hence, the source impedance will be relatively high compared with the majority of SS approaches. A value of around 1k - 3k would be my estimate.
fyi - The cascode device is Bipolar (already have devices riveted to some old heatsinks) and it's base will follow the output signal in order to keep the Vgs of the source follower roughly constant (e.g. at 5V) and at a low value to minimize power dissipation giving me flexibility to use a smaller FET with low input capacitance.
The source follower will have an active source load (again, Bipolar - already have devices riveted to some old heatsinks). The simple approach suggests constant current source, but it will be tempting to play with some other choices such as Aleph / gyrator.
I've never built such an amplifier before and the only building block from those I've just described that I have used before is a common cathode triode, never used a FET before, so part of the fun here is doing something new even if it ain't perfectly engineered
Member
Joined 2009
Paid Member
Now this one looks fun, SiC JFET
I've taken a look at the datasheet. It looks very nice, compared with IRF510 it has a much more convenient TO-247 package for mounting. If I'm reading it right the most linear operating point is at a good 2A - 3A of current.
I'm hoping that used as a Source follower with cascode I won't have to worry too much about input capacitance - it's hard to read the Ciss curves in the datasheet - looks like input capacitance takes a jump at low Vds.
Where does one buy a pair of these ?
I've taken a look at the datasheet. It looks very nice, compared with IRF510 it has a much more convenient TO-247 package for mounting. If I'm reading it right the most linear operating point is at a good 2A - 3A of current.
I'm hoping that used as a Source follower with cascode I won't have to worry too much about input capacitance - it's hard to read the Ciss curves in the datasheet - looks like input capacitance takes a jump at low Vds.
Where does one buy a pair of these ?
Member
Joined 2009
Paid Member
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.