Hi there,
I'm searching for a complementary low power mosfet pair for a VAS stage.
What are available pairs?
2SK216/2SJ79 : are not really electrically complements of eachother according to John Curl, these are good for cascode devices or drivers
2SK2013/2SJ313 : the same as the above pair
ZVP3310/ZVN3310: to low VDS in my case
ZVP4424/ZVN4424: could be usefull, but are they really complementary?
Does anyone know more interesting low power mosfet pairs which are good for a VAS stage?
Thanks
I'm searching for a complementary low power mosfet pair for a VAS stage.
What are available pairs?
2SK216/2SJ79 : are not really electrically complements of eachother according to John Curl, these are good for cascode devices or drivers
2SK2013/2SJ313 : the same as the above pair
ZVP3310/ZVN3310: to low VDS in my case
ZVP4424/ZVN4424: could be usefull, but are they really complementary?
Does anyone know more interesting low power mosfet pairs which are good for a VAS stage?
Thanks
irf540
this one has an Ciss of 870p.
I noticed that Pavel Macura is using mosfet's as VAS devices in his PA4 amplifier. I wonder which types they are?
PA4 power amplifier
this one has an Ciss of 870p.I noticed that Pavel Macura is using mosfet's as VAS devices in his PA4 amplifier. I wonder which types they are?
PA4 power amplifier
Last edited:
Borbely used 2 parrallel and matched 2sk170/2SJ74 as VAS devices with 2SK2013 and 2SJ313 as their cascodes. You get a Jfet/Fet Cascode - in Borbelys case he goes further by differentially driving as Lender and uses the Hawksford cascode - it works very well - listening to it now on my CLX.
Saludos para Buenos Aires
Saludos para Buenos Aires
Last edited:
Indeed, using multiple closely matched Jfet's could be an option.
Ian,
Good question, but I have no idea what the influence would be when N and P are not really complementary. Below is the current design. Can you tell me?
First I used Fet's as drivers but noticed when I'm using them as VAS and BJT't as drivers which load the VAS a bit more, the results are much better.
THD spectrum stays +- the same from 10Hz to 200Khz.
Open loop bandwith is +-10Khz with a Phase Margin of +-85° and Gain margin of 16dB.
Still, it looks strange to me, due to the High Ciss values of a FET, I would expect much higher THD at high frequencies? But leading the drain towards the IPS instead of the rails are keeping Vgs pretty much constant, could this be the reason?
Ian,
Good question, but I have no idea what the influence would be when N and P are not really complementary. Below is the current design. Can you tell me?
First I used Fet's as drivers but noticed when I'm using them as VAS and BJT't as drivers which load the VAS a bit more, the results are much better.
THD spectrum stays +- the same from 10Hz to 200Khz.
Open loop bandwith is +-10Khz with a Phase Margin of +-85° and Gain margin of 16dB.
Still, it looks strange to me, due to the High Ciss values of a FET, I would expect much higher THD at high frequencies? But leading the drain towards the IPS instead of the rails are keeping Vgs pretty much constant, could this be the reason?
Attachments
CAPACITANCE OR TRANSCONDUCTANCE?
In a pre-driver/driver/VAS mosfet can be given requirement to drive the preceeding stage, so that transconductance is a prime parameter, or it is not, so small capacitance is of prime importance (to ensure minimal phase shift and good linearity?).
From reading this site in the past I heard that transconductance per capacitance (e.g. Yfs/Ciss) is the parameter of interest. But my observation with power mosfet didn't conform with this, especially in class-A.
I found that for power mosfet where current drive is always an important requirement, transconductance is a lot more important than capacitance. Even if capacitance is in the order of nF, when the transconductance is high, the sound will be just okay.
But the opposite is true for VAS where current requirement is not critical.
So there is a line of changing requirement from low capacitance small size mosfets to high transconductance power (TO-220) mosfets.
COMPLEMENTARY MATCHING BY DATASHEET
When capacitance is the requirement, then matched capacitance is the key (I currently use Ciss as it represents both capacitances at the drain and source). Transconductance matching is not necessarily required.
When transconductance is the requirement, I prefer to match Rds(on) and of course Yfs/Rds from the datasheet.
The real capacitances and transconductances at the point of interests (current, voltage, frequency, etc.) is unknown, but I think the datasheet parameters can be used for prediction and relative comparison. I even created some matching parameters, which is far from scientific at all
EFFECT OF TRANSCONDUCTANCE VERSUS CAPACITANCE
As driver or output, transconductance is the key. At certain situation capacitance is the key. But what is the formula? How about situation when a bit of transconductance is required as well as small capacitance? Here is the deal...
Transconductance brings up sonic/dynamic, which is important for enjoyment and musicality.
Capacitance will effect something (I guess high frequency distortion, phase inversion and linearity) such that smaller capacitance will bring smoother and clearer sound.
To some extent, this is about taste. As for myself, I have a standard for minimal current drive (transconductance). But don't ask further. Same with capacitance (which requires class-A operation)
In practice, certain threshold numbers can be checked with simulation, and acceptability of transconductance and capacitance amounts is reflected from the choice of parts (I didn't consider the planar type mosfets, which are very good, because I rarely have access to complementary pairs).
CHOICE OF COMPLEMENTARY PAIRS
When transconductance is not required, I found that ZVN3310/ZVP3310 is the best. They have small capacitance and a good match (80%). Max 40pF for ZVN and max 50pF for ZVP.
#2 is BS170 and BS250 which has comparable small capacitance but unpredictable matching. But even tho these are not meant as "complementary" pair, they are widely used in telecom/radio equipments. I use this pair for my projects.
#3 is ZVP2110G/ZVN2110, less than 100pF. Not really complementary (75%). Fetzilla uses only the ZVP so there is no need for matching, just small capacitance.
#4 is VN1210/VP0808, less than 150pF and are very good match (83%). Also has good transconductance (above 0.2S). This is used in JLH89.
#5 is ZVN4424G/ZVP4424G, less than 200pF. Pretty good matching (91%). The transconductance is probably very high but doesn't match well (31%) so only for "VAS".
#6 is K213-6/J76-9 which is typically no more than 120pF. But useless transconductance (too small).
Now when transconductance is critical such as in a driver or preamp output, I consider useful transconductance starts at 0.7S. K2013/J313 is the smallest (typ. 0.7S/210pF), but may need reinforcement and/or paralleling for good current taste, and not really matched (69%).
IRFD120/9120 has waay too high capacitance (450pF).
IRF9610/710, at slightly above 1S and 150pF is actually the best "all round" device, but this also puts them in a difficult position for very demanding listeners (It is either too low transconductance or too high capacitance).
IRF 510/9520 (typ. more than 2S) is prime for transconductance (and is the biggest mosfet suitable for VAS/driver), while maintaining acceptable (to most listener) capacitance typically below 300pF. This can be used in class-A amps (Ayre V3 is an example).
IRF632/9630 is my choice for my TSSA output (typically higher than 3.5S, less than 800pF). For a mixed mosfet-BJT output (BiGBT?) they can also be used (73% matched). Dr Jagodic used IRF630/9630 for DOGC-H, but they are terribly matched imo (36%).
IRF530/9530 has higher transconductance and was used by Dr. Jagodic for DOGC MK3. In my prediction they are terribly matched (39%). I don't know why Dr. Jagodic (Boraomega) switched to IRF630/9630 (equal predicted capacitance) in later DOGC-H. I have built them but never simulated. I guess IRF632/9630 will perform better.
IRF640/9540 is also a good match (93%), but they are from different voltage classes so I guess the numbers are not reliable for comparison. They are still very good match but IRF632/9630 are measured and sound better in my applications.
Attachments
Yes, especially so in a symmetrical topology where you have to pay for any crossover distortion and second harmonic cancellation. This tells why more people prefer to listen to the single ended versions.
Similarly non-complementary. But different function devices. K2013 for transconductance and K216 for capacitance.
The most complementary in my prediction (91%). 3310 is less than 50pF, match 80%. 4424 is less than 200pF, match 91%.
Not useful for drivers. They are suitable only for output stage.
Like DOGC-H? Not a good complement. My TSSA uses 632/9630, for the output stage. The mosfet in the DOGC is more an output than a driver.
Yes, a high current drive, otherwise it will sound terrible that way.
My hypothesis is that we need only to match the capacitance as stated previously.
Yes, except for those who search for extra linearity with minimal correction. I think symmetrical amps must be built with extra care and matching. Otherwise it will only sound boring.
Relatively higher distortion number. And crossover distortion in a symmetrical class-B cannot be "heard" but can be easily "felt" in the long run. Boring, fatiguing, non-musical.
When you put K216/J79 as drivers, they don't have enough transconductance (and their transconductance doesn't match well either). When you use them as VAS you don't need their transconductance but small capacitance (which match quite well). If you need more transconductance for certain circuit, K2013/J313 is a better option.
WOW, thanks Jay!
Very clear information.
After goiing through the Pavel Macura amplifier thread, I already changed the VAS devices for the 2013/313.
PMA is using the power FET's 2SK413/2SJ118 as VAS stage running at 20mA.
http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier.html
Greetz
Very clear information.
After goiing through the Pavel Macura amplifier thread, I already changed the VAS devices for the 2013/313.
PMA is using the power FET's 2SK413/2SJ118 as VAS stage running at 20mA.
http://www.diyaudio.com/forums/solid-state/155628-solid-solid-state-power-amplifier.html
Greetz
Have you read the
http://www.diyaudio.com/forums/soli...vas-lateral-output-perfect.html?highlight=fet
It has lot's of discussion on the differences made by fets cf bjts.
And I tend to agree: exact complementaries are probably not much of an improvement in sound quality.
Have a look at the other Zetex mosFETs and particularly the G version (sot223).
http://www.diyaudio.com/forums/soli...vas-lateral-output-perfect.html?highlight=fet
It has lot's of discussion on the differences made by fets cf bjts.
And I tend to agree: exact complementaries are probably not much of an improvement in sound quality.
Have a look at the other Zetex mosFETs and particularly the G version (sot223).
Ah! Yes you're right! I forgot that DOGC were not push-pull designs so matching is not important. In both amps, the current sources are those with better transconductance (N-channel) which I think is more proper than otherwise.
Moving from 9530 to 9630 was I think a good decision looking from the capacitance (which is important in this position): 860pF to 550pF (Harris/Intersil) typically.
Hi Jay
Your table form post #9 is a great help.
But the 2Sk2013/313 are more complementary than showed in your table.
Have a look at the capture from the Toshibe FET collection.
Both are specified for having a max Rds on of 5Ohm.
Or do you have more data where other values are specified? If Yes, I would love to see it.
I have 20 pairs laying which I would like to use in my current work in progress amp. I will compare them with the 610/9610 when to proto pcb's are ready.
Regards
Your table form post #9 is a great help.
But the 2Sk2013/313 are more complementary than showed in your table.
Have a look at the capture from the Toshibe FET collection.
Both are specified for having a max Rds on of 5Ohm.
Or do you have more data where other values are specified? If Yes, I would love to see it.
I have 20 pairs laying which I would like to use in my current work in progress amp. I will compare them with the 610/9610 when to proto pcb's are ready.
Regards
Attachments
Last edited:
Wonderful complementary DMOS FETs from Supertex
Supertex Low Threshold Enhancement-Mode Vertical DMOS FETs.
Fast.
High drain to source and drain to gate voltage ratings.
Low threshold (2.0V max.).
20 Volt max. gate to source rating.
High input impedance.
Low input capacitance.
Low on-resistance.
No thermal runaway or secondary breakdown.
Low input & output leakage.
Surprisingly high current capability from a small package size.
Available in TO-92 package, perfect for DIY.
(TN2640 / TP2640) - A wonderful complementary pair.
Don't cost a silly furtune and easily obtainable.
Try them, you will really like them.
*TN2640 MODEL
*
.MODEL TN2640 NMOS (LEVEL=3 RS=0.07 NSUB=1.5E15
+DELTA=1.0 KAPPA=1.20 TPG=1 CGDO=1.16E-10
+RD=2.0 VTO=1.20 VMAX=5.0E6 ETA=0.053089
+NFS=6.6E10 TOX=750E-10 LD=1.698E-9 UO=862.425
+XJ=6.4666E-7 THETA=1.0E-5 CGSO=6.13E-10 L=2.5E-6
+W=100E-3 KP=3.5E-6)
*
*TP2640 MODEL
*
.MODEL TP2640 PMOS (LEVEL=3 RS=0.20 NSUB=1.5E15
+DELTA=1.0 KAPPA=1.20 TPG=-1 CGDO=1.916E-10
+RD=9.5 VTO=-1.10 VMAX=5.0E6 ETA=0.053089
+NFS=6.6E10 TOX=750E-10 LD=1.698E-9 UO=862.425
+XJ=6.4666E-7 THETA=1.0E-5 CGSO=1.750E-9 L=2.5E-6
+W=100E-3 KP=2.5E-6)
*
Supertex makes a lot of great, affordable, easily obtainable, semiconductors.
Check out their depletion mode devices and unharness your ideas.
Enjoy.
Sincerely,
Rich B.
Supertex Low Threshold Enhancement-Mode Vertical DMOS FETs.
Fast.
High drain to source and drain to gate voltage ratings.
Low threshold (2.0V max.).
20 Volt max. gate to source rating.
High input impedance.
Low input capacitance.
Low on-resistance.
No thermal runaway or secondary breakdown.
Low input & output leakage.
Surprisingly high current capability from a small package size.
Available in TO-92 package, perfect for DIY.
(TN2640 / TP2640) - A wonderful complementary pair.
Don't cost a silly furtune and easily obtainable.
Try them, you will really like them.
*TN2640 MODEL
*
.MODEL TN2640 NMOS (LEVEL=3 RS=0.07 NSUB=1.5E15
+DELTA=1.0 KAPPA=1.20 TPG=1 CGDO=1.16E-10
+RD=2.0 VTO=1.20 VMAX=5.0E6 ETA=0.053089
+NFS=6.6E10 TOX=750E-10 LD=1.698E-9 UO=862.425
+XJ=6.4666E-7 THETA=1.0E-5 CGSO=6.13E-10 L=2.5E-6
+W=100E-3 KP=3.5E-6)
*
*TP2640 MODEL
*
.MODEL TP2640 PMOS (LEVEL=3 RS=0.20 NSUB=1.5E15
+DELTA=1.0 KAPPA=1.20 TPG=-1 CGDO=1.916E-10
+RD=9.5 VTO=-1.10 VMAX=5.0E6 ETA=0.053089
+NFS=6.6E10 TOX=750E-10 LD=1.698E-9 UO=862.425
+XJ=6.4666E-7 THETA=1.0E-5 CGSO=1.750E-9 L=2.5E-6
+W=100E-3 KP=2.5E-6)
*
Supertex makes a lot of great, affordable, easily obtainable, semiconductors.
Check out their depletion mode devices and unharness your ideas.
Enjoy.
Sincerely,
Rich B.
rich:
thanks for the suggestion.
i'm looking at the curves & specs and wondering what makes them such good complements?
it's also too bad both parts are not available in TO-252 package.
years ago, supertex used to make some great parts that were decent complements in to-220 package that were great for amps and preamps, etc.
thanks for the suggestion.
i'm looking at the curves & specs and wondering what makes them such good complements?
it's also too bad both parts are not available in TO-252 package.
years ago, supertex used to make some great parts that were decent complements in to-220 package that were great for amps and preamps, etc.
Last edited:
I don't remember now. Will check when I can fix my harddisk to get my audio related files including pdfs. Now when I googled I can only see pdf from Toshiba.
I don't remember where I get these 45x and 65x. And I don't remember what does "x" means there. I guess I derived the number indirectly from chart because Rds was not specified.
The 5 Ohm is the Rds max. Note the "MAX".
But that's only tolerance specification. The real value is different issue. And that's just for help. When I have both devices with me (J313 versus IRF610), what matters is what I hear. So you can directly compare them in your specific circuit and with your own ears
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.