Hi everyone,
I have a DIY amp based on 2sk1058 and 2sk162 Hitachi power mosfets. In term of linearity how would it compare with other type of audio power mosfets? Are there better mosfets around?Thanks for your feedback.
Guy🙄
I have a DIY amp based on 2sk1058 and 2sk162 Hitachi power mosfets. In term of linearity how would it compare with other type of audio power mosfets? Are there better mosfets around?Thanks for your feedback.
Guy🙄
It depends on the application. If there were such a thing as
the ideal Mosfet for all linear applications, then everybody
would be using it, and it would become cheap through high
volume.
the ideal Mosfet for all linear applications, then everybody
would be using it, and it would become cheap through high
volume.
If you are talking about output stage MOSFETs I would go for parts with a low gate to source voltage variation related to drain source variations and drain current variations. I suppose that a large MOSFET chip will give you a high Ids/Vgs but I am not sure how Vgs will respond to Vds-variations. The are allways cascodes if you want to keep the voltage drop constant over the transistors😉
That would be a Vertical Mosfet. The other kind is a Lateral,
which has lower transconductance, but also lower Capacitance
Gate-Drain. There are proponents of both types. Charles
Hanson argues for Laterals, I tend to use Verticals, and to
deal with the G-D capacitance, I too would recommend
cascoding.
which has lower transconductance, but also lower Capacitance
Gate-Drain. There are proponents of both types. Charles
Hanson argues for Laterals, I tend to use Verticals, and to
deal with the G-D capacitance, I too would recommend
cascoding.
guymrob said:
I don't know how you would define "better mosfets". There is good news and bad news about these devices, as with most electronic components, and you must make your best judgement call in your design.
These are called "lateral" power MOSFETs, and have much lower transconductance and higher on resistance than typical "vertical" power MOSFETs (that's the bad news). The good news is they can be easily paralleled for higher output, and they need simpler biasing (minimal temperature compensation needed). Unfortunately, they are not well suited for the vast majority of other MOSFET applications, so they have become low volume, high-priced components, compared to vertical MOSFETs.
The advantages of vertical MOSFETs have led them to dominate the market. In fact, there are literally thousands of vertical MOSFETs made, and only a handful of lateral parts available today. The laterals are primarily aimed at the high end audio amplifier market for new design. and replacement market for existing products.
With care in design, the disadvantages of vertical MOSFETs can be largely nullified. Biasing must be temperature compensated. If used in parallel, they must be carefully matched for Vgs(on) and transconductance, and used with large matched source resistors; they should be derated because the sharing will never be perfect.
In your particular design, there are only two other lateral MOSFET device pairs that might be "better": the 2SJ352/2SK2221 pair and the 2SJ201/2SK1530 pair. [I'm ignoring boutique semiconductor vendors, since their parts are hard to get and expensive.] These are rated for slightly higher current and voltage. If you don't need that, I would not bother to change the design.
Actually, the Toshiba 2SJ201/2SK1530 pair are vertical devices with all of the advantages and disadvantages thereof. One nice thing about this pair is that the threshold voltage is lower than found in many vertical devices. Another nice thing is that they are good complements to each other, unlike the hit-and-miss approach you're stuck with for American vertical MOSFETs. (But I'll still stick with laterals.)
The best and the most complementary MOSFETs for audio purposes come from EXICON. Such opinion is delivered by polish manufacturer of High End amps.
http://www.profusionplc.com/product-frame.htm
click EXICON
http://www.profusionplc.com/product-frame.htm
click EXICON
EXICON
Quite interesting...🙂
But the prices are also pretty high....
BTW-has anyone bought the EXICON Audio evaluation board? There was a thread here in 2001 about this, but i don't know if eventually someone bought that.
Regards
Tino
Quite interesting...🙂
But the prices are also pretty high....
BTW-has anyone bought the EXICON Audio evaluation board? There was a thread here in 2001 about this, but i don't know if eventually someone bought that.
Regards
Tino
Re: EXICON
Exicon expensive??? Maybe there are countries where you
can buy the Hitachi devices much cheaper, but here in Sweden
at least they cost around three times the price of the Exicons.
(Yes, one has to order Exicon from the UK, but since they don't
charge for shipping, have no minimum order size and there are
no customs fees within EU, it is fair to compare with swedish
off-the-shelf prices). I have seen lower prices for the Hitachis
elsewhere, but they are still more expensive than Exicon.
Further, the Exicons are slightly higher rated for max Id.
zinsula said:
But the prices are also pretty high....
Exicon expensive??? Maybe there are countries where you
can buy the Hitachi devices much cheaper, but here in Sweden
at least they cost around three times the price of the Exicons.
(Yes, one has to order Exicon from the UK, but since they don't
charge for shipping, have no minimum order size and there are
no customs fees within EU, it is fair to compare with swedish
off-the-shelf prices). I have seen lower prices for the Hitachis
elsewhere, but they are still more expensive than Exicon.
Further, the Exicons are slightly higher rated for max Id.
Exicon vs Hitachi
You can get 2SK1058 in Germany for under 6 Euro (= 4 GBP). I know at least 2 sources, one of which is Reichelt.
And if you look at the datasheet from Exicon, I wouldn't consider them linear at all, especially if you bias them at around 1 to 1.5A. They might be a bit better at 3A, but then you have to cascode to get the heat down.
Compare to the few 2SK1058's and the 2SK1529's that I have measured, I'll rather stick to the Japanese. But if you can live with second harmonics (e.g. drive them 180 deg out of phase like in JLH 10W), IRFP is unbeatable for price.
Patrick
You can get 2SK1058 in Germany for under 6 Euro (= 4 GBP). I know at least 2 sources, one of which is Reichelt.
And if you look at the datasheet from Exicon, I wouldn't consider them linear at all, especially if you bias them at around 1 to 1.5A. They might be a bit better at 3A, but then you have to cascode to get the heat down.
Compare to the few 2SK1058's and the 2SK1529's that I have measured, I'll rather stick to the Japanese. But if you can live with second harmonics (e.g. drive them 180 deg out of phase like in JLH 10W), IRFP is unbeatable for price.
Patrick
Re: Exicon vs Hitachi
There seems to be some confusion on this point. FETs are supposed to have a curved transfer function that follows the "square law". That is, the drain current is supposed to vary as the square of the gate voltage.
The Exicon data sheet shows this happening at lower currents. At higher currents the curve bends the other way. This is due to the channel resistance increasing at high temperatures. Somewhere in the middle, these two effects partially cancel, giving a relatively straight line. But this would only result in a more "linear" amplifier if you were using it in a common-source configuration with no source degeneration resistor.
The original poster is using these devices in a common-drain configuration. In his situation, he mostly wants to increase the overall transconductance to increase the linearity. The easiest way for him to do this is to use the Magnatec "double die" parts. These are essentially two transistors in one package. In this way he can double the transconductance with no other circuit alterations required.
EUVL said:
There seems to be some confusion on this point. FETs are supposed to have a curved transfer function that follows the "square law". That is, the drain current is supposed to vary as the square of the gate voltage.
The Exicon data sheet shows this happening at lower currents. At higher currents the curve bends the other way. This is due to the channel resistance increasing at high temperatures. Somewhere in the middle, these two effects partially cancel, giving a relatively straight line. But this would only result in a more "linear" amplifier if you were using it in a common-source configuration with no source degeneration resistor.
The original poster is using these devices in a common-drain configuration. In his situation, he mostly wants to increase the overall transconductance to increase the linearity. The easiest way for him to do this is to use the Magnatec "double die" parts. These are essentially two transistors in one package. In this way he can double the transconductance with no other circuit alterations required.
Looking at Magnatec's and EXICON's datasheets I conclude all parameters of complementary lateral MOSFETs for audio purposes are the same (excluding an evident error in Magnatec's MAG90X95 and MAG91X96 specification; ton for P channel MOSTEF is printed 12 ns instead of 120 ns). It means that semiconductor structure comes from Magnatec's BUZ9xx MOSFET series and differences are due to encapsulation process.
What is the bias current for Magnatec BUZ900DP & BUZ905DP?
Hi Charles and everyone,
I got a couple of Magnatec BUZ900DP and BUZ905DP to try out, looks like they are bigger than the Hitachi ones. I managed to have a nice fit into the heatsink. It sounded tighter, esp. the bass, i.e. more controlled at the bottom. The mid remaind a bit warm while the high is sweet and extended. I like it!
The previous bias current for Hitachi 2sk1058 & 2sk162 is 100mA, what is the optimium bias current for Magnatec BUZ900DP & BUZ905DP? Thanks for the inputs!
Regards,
Guy
Attached is the pic of my DIY amplifier.
Hi Charles and everyone,
I got a couple of Magnatec BUZ900DP and BUZ905DP to try out, looks like they are bigger than the Hitachi ones. I managed to have a nice fit into the heatsink. It sounded tighter, esp. the bass, i.e. more controlled at the bottom. The mid remaind a bit warm while the high is sweet and extended. I like it!
The previous bias current for Hitachi 2sk1058 & 2sk162 is 100mA, what is the optimium bias current for Magnatec BUZ900DP & BUZ905DP? Thanks for the inputs!
Regards,
Guy
Attached is the pic of my DIY amplifier.
Attachments
Someone told me here once that logic level mosfets have a higher transconductance than a similarly rated "normal" one. Also, the rated transconductance for a given mosfet may be at a current way higher than what you are going to use it at. Have a look at the published curve.Charles Hansen said:
Re: What is the bias current for Magnatec BUZ900DP & BUZ905DP?
The Hitachi are packaged in what is called TO-247 in the US while the BUZ90XDP are packaged in TO-264, which is slightly larger. I'm glad that you found the improvement to your liking, and somewhat surprised you were able to pick out the useful advice buried in this less-than-focused thread. 🙂
The zero-tempco bias for the double-die parts is around 240 mA. However, I'm not sure if your heatsink is big enough to handle that much power. The best rule of thumb is to make sure the heatsinks don't go much past 130 F or so. This is partly for the safety of the devices, but also for the safety of anyone that might accidentally touch the heatsinks. 130 F is easily judged without a thermometer -- it is when the heatsink feels uncomfortably (not painfully!) hot after resting your hand on it for 3 seconds or so.
Don't worry if your heatsinks won't allow for this much bias. Just make sure the fins are vertically oriented and bias them up until the heatsink temperature is around 130 F.
guymrob said:
The Hitachi are packaged in what is called TO-247 in the US while the BUZ90XDP are packaged in TO-264, which is slightly larger. I'm glad that you found the improvement to your liking, and somewhat surprised you were able to pick out the useful advice buried in this less-than-focused thread. 🙂
The zero-tempco bias for the double-die parts is around 240 mA. However, I'm not sure if your heatsink is big enough to handle that much power. The best rule of thumb is to make sure the heatsinks don't go much past 130 F or so. This is partly for the safety of the devices, but also for the safety of anyone that might accidentally touch the heatsinks. 130 F is easily judged without a thermometer -- it is when the heatsink feels uncomfortably (not painfully!) hot after resting your hand on it for 3 seconds or so.
Don't worry if your heatsinks won't allow for this much bias. Just make sure the fins are vertically oriented and bias them up until the heatsink temperature is around 130 F.
Oscillation mosfet 2SK1058
Hello,
I have registered very interested in your conversation of the mosfet. I have some oscillation inconveniences in my amplifier with the 2SK1058 2SJ162, I have solved the problem to increasing the Bias
But I am not for sure this it is the best way to avoid the oscillation of the mosfet of power.
Probably, you can help me to find a better solution.
Also, I have the restlessness and necessity of finding a more better alternative for the 2SK 1058. Somebody has experienced with another model of Mosfet like the 2SK 1529 and he obtained better result. ?
Hello,
I have registered very interested in your conversation of the mosfet. I have some oscillation inconveniences in my amplifier with the 2SK1058 2SJ162, I have solved the problem to increasing the Bias
But I am not for sure this it is the best way to avoid the oscillation of the mosfet of power.
Probably, you can help me to find a better solution.
Also, I have the restlessness and necessity of finding a more better alternative for the 2SK 1058. Somebody has experienced with another model of Mosfet like the 2SK 1529 and he obtained better result. ?
I think the 1529 and that family have a much larger Vgs than the lateral 1058. It might be a vertical FET.
It will need a re-design of the bias system and care in setting up the temperature compensation.
It is a good robust replacement if you take Gds and positive tempco. into account.
re. your oscillation there are others more experienced to help with this but I do recommend that you do not increase the value of the Miller compensation. Find another suppression route to suit your PCB layout.
It will need a re-design of the bias system and care in setting up the temperature compensation.
It is a good robust replacement if you take Gds and positive tempco. into account.
re. your oscillation there are others more experienced to help with this but I do recommend that you do not increase the value of the Miller compensation. Find another suppression route to suit your PCB layout.
Re: Oscillation mosfet 2SK1058
One more thing is to check is you get oscillations when the output current is above a certain level, like 1 A or so. Check especially the N-channel device (= postive signals).
If you have oscillation problems you could increase the gate resistors a bit but be careful not increasing too much becasue the transistors get slower also.luclucmir said:
One more thing is to check is you get oscillations when the output current is above a certain level, like 1 A or so. Check especially the N-channel device (= postive signals).
Re: Oscillation mosfet 2SK1058
Make sure you are using gate "stopper" resistors as close to the gates as physically possible. As a starting point, these should be somewhere in the range of a few hundred ohms to a few thousand ohms.
Assuming that you are using source followers in the output stage, another point that can help with oscillations is to run a small cap (1000 pF or so) from the drain of each device to the heatsink itself. This idea was taken from the Hitachi application sheet, and will stop oscillations in many cases. It is essentially a local bypass cap on the power supply rails to the closest chassis ground, so it won't hurt the sonics at all.
Good luck!
luclucmir said:
Make sure you are using gate "stopper" resistors as close to the gates as physically possible. As a starting point, these should be somewhere in the range of a few hundred ohms to a few thousand ohms.
Assuming that you are using source followers in the output stage, another point that can help with oscillations is to run a small cap (1000 pF or so) from the drain of each device to the heatsink itself. This idea was taken from the Hitachi application sheet, and will stop oscillations in many cases. It is essentially a local bypass cap on the power supply rails to the closest chassis ground, so it won't hurt the sonics at all.
Good luck!
If you are concerned with linearising your MOSFETs AND allaying some temp/bias vagaries, why not do a CFP ? A quite low power driver like 2n5551/5401 would ensure low Vas loading c.f. HF gate capacitance drive.
- Status
- Not open for further replies.