It's an Andrew issue - I posted a 46k *gif file. There's nothing to fix. Your complaining will accomplish nothing but irritating everyone else on this thread.
Google Chrome rules. Quicker and more streamlined than IE and FF. Try it Andrew. I think you may like it.
OK, some more dumb queries:
PN439x series is not always available here without placing large parts orders, cost etc.
I understand the "x" is same as in J111 for PN4391 and the characteristics are cross-
referenced in some supplier cats. Is this so and applying also to J112,3? If OK, I presume
straight substitutions might be acceptable....or not?
Google Chrome?...Yep, fine for me with XP !
PN439x series is not always available here without placing large parts orders, cost etc.
I understand the "x" is same as in J111 for PN4391 and the characteristics are cross-
referenced in some supplier cats. Is this so and applying also to J112,3? If OK, I presume
straight substitutions might be acceptable....or not?
Google Chrome?...Yep, fine for me with XP !
I did some looking around, and at least in the Siliconix catalog, the J111 series and the PN4391 series come from different process families (NC vs. NCB), so they aren't necessarily a real cross. This is not to say that the J111 won't work in place of the 4391, but it's probably going to act a bit differently. You might try looking for the PN4091-4093 series. I tried simulating using the 4093 in place of a PN4393 in one circuit yesterday, and the result was ok. The PN4091 and PN4391 series (at least when made by Siliconix), came from the same process family.
Thanks for your assistance, I also sorted through Fairchild's JFET clones and found that all their switching N types are process 51.
That's PN409x, PN439x, J11x, 2N5639 and probably others too. They seem to go in threes everywhere through the number systems. GP amplifiers and RF types are likewise standardised to just 1 or 2 processes each so I guess their process capabilities were considered good enough to cover all types. Makes you wonder about what you really have as a product, though 😕
Still, they seem to produce the widest range and hold useful stocks of JFETs.
I don't think we need dispute their specs., so they will likely be the manufacturer for many builders, me included.
Last edited:
Fairchild has been sucking up the small-signal product lines of quite a few companies that have decided to kill those lines. They have been also doing this in the case of optoelectronic parts like optoisolators. When they have been able to get away with it (such as the cases where the part number is not JEDEC registered), they have played with the specs, probably to make the consolidation you speak of possible.
I agree with you - for many people, Fairchild will probably be the only game in town, as the number of players in the jfet business has shrunk alarmingly. I have bought their parts to try them out (they are quite cheap), but have been reluctant to use them extensively.
As to the suitability of the J111 for this application, I'll have to do a simulation to see. If I recall, the IDSS specs for that series are on the low side compared to the PN4391 series, especially in the case of the J113. This may not matter all that much as you can cherry-pick in a bunch of parts to find what you need. It will make a difference in the simulation, as the cutoff voltage for the J111 model will likely be lower that that of the 4391. The part will work anyway if you bias it right. The distortion will probably be higher, but likely still quite good.
I have another option for a lineamp with gain that I plan to try in the next few months. The distortion is about the same or a little less than the circuit shown in this thread, though it probably won't be able to source as much current unless the output buffer is really beefed up. That may not be a really big deal - I've never understood the people that think you need a line amp capable of driving a toaster. The new amp is all jfet, though it requires a p-channel device.
I agree with you - for many people, Fairchild will probably be the only game in town, as the number of players in the jfet business has shrunk alarmingly. I have bought their parts to try them out (they are quite cheap), but have been reluctant to use them extensively.
As to the suitability of the J111 for this application, I'll have to do a simulation to see. If I recall, the IDSS specs for that series are on the low side compared to the PN4391 series, especially in the case of the J113. This may not matter all that much as you can cherry-pick in a bunch of parts to find what you need. It will make a difference in the simulation, as the cutoff voltage for the J111 model will likely be lower that that of the 4391. The part will work anyway if you bias it right. The distortion will probably be higher, but likely still quite good.
I have another option for a lineamp with gain that I plan to try in the next few months. The distortion is about the same or a little less than the circuit shown in this thread, though it probably won't be able to source as much current unless the output buffer is really beefed up. That may not be a really big deal - I've never understood the people that think you need a line amp capable of driving a toaster. The new amp is all jfet, though it requires a p-channel device.
I've been wrestling with the J111 in a simulation, and it indeed will need a lower bias current to center the output voltage properly, resulting in higher distortion (I'm not entirely done yet, but that appears to be the general trend). Electronic Goldmine is selling NOS National PN4391 at reasonable prices, and I noticed that they do ship to Oz. This may be an alternative. A padded postal envelope may not be all that expensive. Avoid UPS for international shipping.
These are the results I've obtained so far with the J111. It needs lower bias current as dictated by the model in PSpice, so the distortion is higher. One could possibly cherry-pick devices with higher IDSS /cutoff voltage to fix this. Knowing Fairchild, they might take the devices with higher IDSS in that family and call them PN4391s, leaving the also-rans for J111 duty. It's hard to say.
Attachments
Thanks again for that work, it's instructive to see the circuit operation with different devices. It seems that to do the design justice, the PN4391, among all the available low-cost TO92 parts, is the only suitable one. 'Seems 100-off prices are around 25c US so I'd better get an order together and pursue this properly.
There are quite a few other semis but most seem common, low cost types. Do you have any comment about selection of the VFET regs (IRF612 doesn't seem available). Will any complementary low power TO220 VFETs work there? 2SA1220 and PN4303 are unusual, but I guess every DIY has a stash of 2SJ74s somewhere. Are the FETs both just there as GP amps?
My apoloogies to others reading the thread and finding a bit of a parts-fest. However, I hope we'll also get a feel for circuit operation this way. Thanks for your patience. 🙂
The PN4091 looks like it would also be a suitable substitute for the PN4391. The PN4391 is more common, at least in the US. If you were lucky enough to happen on an NOS stash of the original TO-18 part, the 2N4391, these would work as well.
I had a close look at the schematic, and there are a couple of mistakes. I willl clean it up and repost.
The KSA1220 is a TO-126 case fast bipolar suitable for power amp VAS duty. Comparably fast devices with a VCEO of >40V would work. The KSAs are cheap, and in stock at Mouser. I use them for VAS stages in my power amp experiments. I used the same part number for both parts of the discrete darlington so that I could mount both transistors on a common heat sink. It may also be possible to use a device in the TO-92L case or one of the Zetex E-line package devices and ditch the heat sink. I'll have to stare at the simulation and calculate the power dissipation to be sure.
In the power supply, I used the IRF612 because I have a couple of tubes of them. IRF610 or 510 would work just fine. An IRF9510 could be subbed for the IRF9610 - again, I used what I had on hand. The PN4303 in the power supply is nothing more than a current source set up for ~1mA. A 2N5457 or J202 could also work. In the case of the 2N5457, you'd need to increase the value of the dropper resistor in series with its drain (R25) to 20k or so.
I'll post more comments about the circuit as time allows. The most important will be how to select the jfets used in this circuit, as you can't just sling the preamp together from a set schematic and expect it to work, unless you get lucky - all the jfets need to be selected, and some resistor values may need to be tweaked depending on the results of the selection.
Unfortunately, this is one of those "advanced" projects...
I had a close look at the schematic, and there are a couple of mistakes. I willl clean it up and repost.
The KSA1220 is a TO-126 case fast bipolar suitable for power amp VAS duty. Comparably fast devices with a VCEO of >40V would work. The KSAs are cheap, and in stock at Mouser. I use them for VAS stages in my power amp experiments. I used the same part number for both parts of the discrete darlington so that I could mount both transistors on a common heat sink. It may also be possible to use a device in the TO-92L case or one of the Zetex E-line package devices and ditch the heat sink. I'll have to stare at the simulation and calculate the power dissipation to be sure.
In the power supply, I used the IRF612 because I have a couple of tubes of them. IRF610 or 510 would work just fine. An IRF9510 could be subbed for the IRF9610 - again, I used what I had on hand. The PN4303 in the power supply is nothing more than a current source set up for ~1mA. A 2N5457 or J202 could also work. In the case of the 2N5457, you'd need to increase the value of the dropper resistor in series with its drain (R25) to 20k or so.
I'll post more comments about the circuit as time allows. The most important will be how to select the jfets used in this circuit, as you can't just sling the preamp together from a set schematic and expect it to work, unless you get lucky - all the jfets need to be selected, and some resistor values may need to be tweaked depending on the results of the selection.
Unfortunately, this is one of those "advanced" projects...
I don't know precisely what you're looking for (the link does nothing for me), but if you load up Google with "Hood Liniac", you get a Paul Kemble page with lots of interesting Hood circuits and derivatives, and further down the line, a scan of one of the original Hood Liniac articles from Wireless World. My implementation is not really what Hood was touting in those early articles, though I definitely stole some ideas from him.
Anyway, a lot of the applications of the original Liniac seem to be aimed at higher gain - this is relatively easy. Unity gain (or close enough so that the difference is purely academic) is easy. Getting a gain of 5-10X with low distortion and a single-ended circuit is the rub. So far, this circuit is the best I can come up with to meet that goal without going to Japanese complementary fets (these may still not do it, I need to do more simulations) or caving in and using an opamp. I will post more developments as they happen. Significant diversions from the circuit described here will get separate threads.
What you see in this thread started out as a "what if" kind of deal. I have several circuits that used a fairly hotly biased small-sized mosfet (ZVN3306 (Zetex/Diodes, Inc.), VN0106N3 (Supertex), and MPF960 (Motorola/On Semiconductor) are favorites) for unity/medium gain with relatively low distortion. Since the mosfets in question are enhancement mode, bias is somewhat of a PITA, though reasonably straightforward if you look at it hard enough. Anyway, I use PN4391 jfets (high IDSS, high pinch-off voltage) all the time for cascoding jfets with lower pinch-off voltages, and I mused to myself about what I could get away with if I tried to use one in a low-gain line amp. After some simulation slogging, the result was the circuit presented here. The resulting circuit sounds very nice in my living room setup, and the extra gain (I was using a unity gain line amp prior to this) really peps up my system.
Good luck in your search.
Anyway, a lot of the applications of the original Liniac seem to be aimed at higher gain - this is relatively easy. Unity gain (or close enough so that the difference is purely academic) is easy. Getting a gain of 5-10X with low distortion and a single-ended circuit is the rub. So far, this circuit is the best I can come up with to meet that goal without going to Japanese complementary fets (these may still not do it, I need to do more simulations) or caving in and using an opamp. I will post more developments as they happen. Significant diversions from the circuit described here will get separate threads.
What you see in this thread started out as a "what if" kind of deal. I have several circuits that used a fairly hotly biased small-sized mosfet (ZVN3306 (Zetex/Diodes, Inc.), VN0106N3 (Supertex), and MPF960 (Motorola/On Semiconductor) are favorites) for unity/medium gain with relatively low distortion. Since the mosfets in question are enhancement mode, bias is somewhat of a PITA, though reasonably straightforward if you look at it hard enough. Anyway, I use PN4391 jfets (high IDSS, high pinch-off voltage) all the time for cascoding jfets with lower pinch-off voltages, and I mused to myself about what I could get away with if I tried to use one in a low-gain line amp. After some simulation slogging, the result was the circuit presented here. The resulting circuit sounds very nice in my living room setup, and the extra gain (I was using a unity gain line amp prior to this) really peps up my system.
Good luck in your search.
Last edited:
Thanks for your post. I wanted to read that pdf file and see what it has to offer. Not too important though. I will search for it as you suggested.
Just out of curiosity. Have you ever tried the split supply version of the JLH headphone amp as a preamp ? If so , how does it compare with your current preamp.
Cheers,
Ashok.
Just out of curiosity. Have you ever tried the split supply version of the JLH headphone amp as a preamp ? If so , how does it compare with your current preamp.
Cheers,
Ashok.
Actually. for the past 3-4 years I've been limiting myself to a single +30V supply, derived from an external +40V supply. This is 1) out of laziness 2) for the challenge. There a couple of split supply solutions I might try, but I don't think I've exhausted the single supply options as of yet.
'Seem like good motivations to me. We are awash with thousands of opamp or discrete
opamp designs in most low-level applications. There is something elegant about a design
that seems to grow around its task rather than be squeezed down to fit it, if I can put it
in that impressionistic sense. So it's a complete design rather than just an integrated
device application.
Maybe it's old-tech and sub-optimal to conventional audio designers but there are some
advantages to subjective sound with single rail supplies and non-symmetric topology too,
regardless of how low the THD can be pushed, IME.
I think you benefit us all by sticking at it. Great stuff, Wrenchone.
opamp designs in most low-level applications. There is something elegant about a design
that seems to grow around its task rather than be squeezed down to fit it, if I can put it
in that impressionistic sense. So it's a complete design rather than just an integrated
device application.
Maybe it's old-tech and sub-optimal to conventional audio designers but there are some
advantages to subjective sound with single rail supplies and non-symmetric topology too,
regardless of how low the THD can be pushed, IME.
I think you benefit us all by sticking at it. Great stuff, Wrenchone.
I think single supply does have it's own characteristics that could make it sonically different. First of all it doesn't have a second supply and the associated PSRR . Then it has an output cap that could be removed if one relies on the input cap of the power amp which most likely is an expensive film type. This capacitor ( on the power amp input or the preamp output ) will have dc bias on it. This I think is good for the sound path . DC biased caps sound better than no bias elcos or even non-polar elcos.
My single supply and split supply JLH headphone amps sound slightly different from each other with mids appearing to be better on the single supply version.
On headphones the output cap doesn't kill the bass quality much though the direct connection on the split supply version does have a very slight improvement in 'taut' bass sounds. However this could possibly improve with a better output capacitor. I used a garden variety Samwha cap.
Then there is the HUGE advantage that an accidental dc off set cannot kill a $300/- headphone in a jiffy ! Just for that it's worth using the single supply version with an expensive output cap. But then as I said earlier , even a low cost one sounds very good.
As a preamp the single supply version is very easy to deal with.
About the extra gain. I found preamps with some gain always sounded better than just a buffer even when dealing with DAC's with a 2 volt output.
My single supply and split supply JLH headphone amps sound slightly different from each other with mids appearing to be better on the single supply version.
On headphones the output cap doesn't kill the bass quality much though the direct connection on the split supply version does have a very slight improvement in 'taut' bass sounds. However this could possibly improve with a better output capacitor. I used a garden variety Samwha cap.
Then there is the HUGE advantage that an accidental dc off set cannot kill a $300/- headphone in a jiffy ! Just for that it's worth using the single supply version with an expensive output cap. But then as I said earlier , even a low cost one sounds very good.
As a preamp the single supply version is very easy to deal with.
About the extra gain. I found preamps with some gain always sounded better than just a buffer even when dealing with DAC's with a 2 volt output.
Last edited:
- Status
- Not open for further replies.