Referring to Patrick and Steen's Mark 2 headphone amp...
--I would think that 10k at the input would be sufficient. Possibly even lower still if your source/preamp can handle it.
--Yes, the follower headphone amp has a very low output impedance. I'm not sure I buy into arguments that headphones need such high output impedances, but in any event it's easy enough to put any series value at the output that you might want to try...even a pot so you can dial it in. Go for it.
--Although I'm happy to send people as many Lovoltech JFETs as they want, I don't think that you'll get much benefit from using one in the current source. It would be operating under DC conditions, hence invisible sonically. Any sonic imprint would be determined by the MOSFET.
--I'm still suspicious of DC at the output. Yes, I'm sure you can get it squared away at idle, but thermal drift is not likely to be so kind.
Carry on.
Grey
--I would think that 10k at the input would be sufficient. Possibly even lower still if your source/preamp can handle it.
--Yes, the follower headphone amp has a very low output impedance. I'm not sure I buy into arguments that headphones need such high output impedances, but in any event it's easy enough to put any series value at the output that you might want to try...even a pot so you can dial it in. Go for it.
--Although I'm happy to send people as many Lovoltech JFETs as they want, I don't think that you'll get much benefit from using one in the current source. It would be operating under DC conditions, hence invisible sonically. Any sonic imprint would be determined by the MOSFET.
--I'm still suspicious of DC at the output. Yes, I'm sure you can get it squared away at idle, but thermal drift is not likely to be so kind.
Carry on.
Grey
> I'm still suspicious of DC at the output. Yes, I'm sure you can get it squared away at idle, but thermal drift is not likely to be so kind.
That is why I duplicated the Loveltech with cascode on the bottom side. Sure, a DN2540 will be good enough as a current source, but a matched Loveltech under the same operating conditions AND thermally tightly coupled to the upper Loveltech will have the same thermal drift, hence the chance of maintaining DC. This is just copied from the Curl JFET follower, not my invention.
i.e. the lower Loveltech is there to compensate for DCdrift, not for sonics. A bit of a waste, I know. But a coupling cap will probably cost much more, and imposes its own sonic signature.
And the lower the output impedance, the less the amount of any DC voltage. Extra motivation to go low.
If it proves to be not stable enough, we can always put the cap back in and swap the lower FETs for a DN2540.
Only experiment will tell.
Patrick
That is why I duplicated the Loveltech with cascode on the bottom side. Sure, a DN2540 will be good enough as a current source, but a matched Loveltech under the same operating conditions AND thermally tightly coupled to the upper Loveltech will have the same thermal drift, hence the chance of maintaining DC. This is just copied from the Curl JFET follower, not my invention.
i.e. the lower Loveltech is there to compensate for DCdrift, not for sonics. A bit of a waste, I know. But a coupling cap will probably cost much more, and imposes its own sonic signature.
And the lower the output impedance, the less the amount of any DC voltage. Extra motivation to go low.
If it proves to be not stable enough, we can always put the cap back in and swap the lower FETs for a DN2540.
Only experiment will tell.
Patrick
GRollins said:
I have given my Diamond HeadPhone Buffer
same conditions as Grey MOSFET:
+/-25 Volt and bias 500 mA
You can use a 2x18 VAC transformer for my amp
and skip 10 Volts drop in rectifier. Save you some money.
(But you may keep 2x 50.000uF supply filtering caps, if you think it is worth it
It will not destroy or improve very much.
Use a small, 0.1 Ohm power resistor, to stop inrush currents
especially if you use a TOROID transformer.)
My own opinion is, for my Diamond Buffer, you can do well
with using 4.700 - 10.000uF. It would do fine to filter only 0.5A.
And here you wont need any small resistor before Caps.
I have spent like 4-5 hours
- setting this circuit up
- do simulations, THD + AC Analysis + Fourier tests at all levels
- alter and test again
- change components values and test again.
In the span of 1 - 30 milliWatt into 32 Ohm:
< 0.004% THD distortion
Nice falling Fourier harmonics sequence
with 3rd harmonics 20-30 dB below 2nd harmonics
at all levels ( 1-30mW ).
You think 30 mW is very little?It is not!
for a pair of HeadPhones 32 Ohm, it is common they have
like 90 dB SPL ( sound pressure level ) at 1 mW.
10mW is +10 dB from this ...
30mW is +15 dB ........ 90+15 = 105 dB
If you try this at home, you risk to get tinnitus very quickly
and your hearing may be damaged, for Rest of your LIFE
Don't go there, please I myself, Lineup, suffer from Tinnitus off and on.
I do not wish anybody to have the same problem....'
Regards
lineup
Attachments
Here is the mockup It played music right away, which is always nice I had all the components lying so it didnt take long to solder the amp together. The low partscount lends itself well to P2P wiring. I didnt have 3r powerresistors so I put a 1r in series with a 1,8r, just in case anyone wonders from the pics. The heatsinking should be covered, dont you think
As you can see on the pic, the DC at the output sits at approx 35mV (negative) and is pretty constant. The value only varies a few mV.
The temp on the IR thermometer is measured directly on the mosfets.
With no signal the amp is deadsilent, actually it must be the most silent headphone amp I have heard. With the preamp at full throttle, there is only a very slight hiss which I am sure comes from the preamp. BTW it is running from the NS10 clone.
I didnt implement the output resistor yet, I will try that later. I will also look into getting the DC tweaked closer to 0V. My idea is to solder a trimpot across one of the source resistors and see if I can get rid of it.
Steen
(More pics comming)
It played music right away, which is always nice I had all the components lying so it didnt take long to solder the amp together. The low partscount lends itself well to P2P wiring. I didnt have 3r powerresistors so I put a 1r in series with a 1,8r, just in case anyone wonders from the pics. The heatsinking should be covered, dont you think As you can see on the pic, the DC at the output sits at approx 35mV (negative) and is pretty constant. The value only varies a few mV.
The temp on the IR thermometer is measured directly on the mosfets.
With no signal the amp is deadsilent, actually it must be the most silent headphone amp I have heard. With the preamp at full throttle, there is only a very slight hiss which I am sure comes from the preamp. BTW it is running from the NS10 clone.
I didnt implement the output resistor yet, I will try that later. I will also look into getting the DC tweaked closer to 0V. My idea is to solder a trimpot across one of the source resistors and see if I can get rid of it.
Steen
(More pics comming)
Attachments
The sound is extremely detailed and direct without getting annoying at any point. It sounds crystal clear and very good indeed. The mini-Zenamp sounds a bit more romantic in comparison. I will do a A/B test later to better give an account of it.
The amp has plenty of power for my Beyerdynamics DT880's, which sounds very, very good with EUVL's amp. The detailed sound is some of the best I have heard yet. Details are just clearly cut out. On many records, they add some artificial soundstage and eccho's and so on. If the technician didnt do a good job, EUVL's amp will tell you through a good pair of cans.
I found myself laughing a couple of times, for that reason
Thanks for posting the circuit, Patrick This amp is absolutely a keeper
Here is a close up of one channel.
Steen
The amp has plenty of power for my Beyerdynamics DT880's, which sounds very, very good with EUVL's amp. The detailed sound is some of the best I have heard yet. Details are just clearly cut out. On many records, they add some artificial soundstage and eccho's and so on. If the technician didnt do a good job, EUVL's amp will tell you through a good pair of cans.
I found myself laughing a couple of times, for that reason
Thanks for posting the circuit, Patrick
This amp is absolutely a keeperHere is a close up of one channel.
Steen
Attachments
The supply is a regulated Zen, split supply. The trafo I had on hand was 2x30V so the easy way was to build a regulator for it.
The regulator is putting out a steady +/- 24V.
Its very clean, since there isnt even the slightest trace of hum. There is nothing worse than hum, in a pair of headphones.
Steen
The regulator is putting out a steady +/- 24V.
Its very clean, since there isnt even the slightest trace of hum. There is nothing worse than hum, in a pair of headphones.
Steen
Attachments
Steen,
One happy customer !!
A few remarks :
a) I would encourage you to trim the DC down to zero by changing the value of the source resistor of the lower FET.
b) I would suggest putting the two loveltechs as close to each other as possible to enhance thermal coupling.
c) The Zen regulated supply is fine. But you might want to try some Japanese FETs just for fun. I would suggest 2SK2955 or 2SK2586 for the regulator.
And if you are now hooked on Loveltechs, send me a private mail. I have something else in the drawer.
And maybe someone would start making PCB's ??
Patrick
One happy customer !!
A few remarks :
a) I would encourage you to trim the DC down to zero by changing the value of the source resistor of the lower FET.
b) I would suggest putting the two loveltechs as close to each other as possible to enhance thermal coupling.
c) The Zen regulated supply is fine. But you might want to try some Japanese FETs just for fun. I would suggest 2SK2955 or 2SK2586 for the regulator.
And if you are now hooked on Loveltechs, send me a private mail. I have something else in the drawer.
And maybe someone would start making PCB's ??
Patrick
Absolutely
Answers to abc:
a: That is my intention. I will look into it tomorrow, allthough there is no problems that I noticed from the DC offfset. Knowing that it excists is reason enough to do something about it, though
b: In the final build, I will optimise the layout, quite a bit. Since this is in fact a quick mockup, I just used the threaded holes allready made in the heatsinks
c: I dont know; the irfp240 and 9240 does a good job. Since I dont have any of the japanese fets, I am a bit reluctant to change them. In the final build, I will go dual mono. I will also use a trafo with lower sec. voltage. The heatsink with the regs get hot
since the reg is throwing off nearly 20 volts!
I said earlier that the heatsinking was taken care off! Well it is, but I wouldnt want to use anything smaller than those on the pics. They reach more than a handwarm temp when the amp has been on for a few hours.
The detailed sound I was talking about earlier, is only possible with a good pair of headphones and an amp to match them. This Lovoltech thing does just that
You wont hear those things on a pair of speakers, by a wide margin Steen
I could not resist the temptation, so this is the first shot at a PCB.
I also gave the circuit a name -- DAO JFET Follower.
It is single sided, to be mounted to the heatsink with a single M2.5x5 standoff, with the Loveltechs and the MOSFETs directly mounted onto the heatsink (mica / grease). The standoff can also act as a single star ground for power / input / output.
All other components have to be soldered first, and have all excess leads trim to minimum (less than 2mm protruding from PCB). The 4 transistors have to be soldered using a little fixture in order to have access to the solder pads, before the entire assembly is mounted to the heatsink.
If someone is interested in converting the layout from Autocad to Gerber, he may contact me.
Patrick
PS I did not quite keep the promise of mounting the Loveltechs side by side. But they have such short leads that such a layout would require a separate heatsink for the JFETs.
I also gave the circuit a name -- DAO JFET Follower.
It is single sided, to be mounted to the heatsink with a single M2.5x5 standoff, with the Loveltechs and the MOSFETs directly mounted onto the heatsink (mica / grease). The standoff can also act as a single star ground for power / input / output.
All other components have to be soldered first, and have all excess leads trim to minimum (less than 2mm protruding from PCB). The 4 transistors have to be soldered using a little fixture in order to have access to the solder pads, before the entire assembly is mounted to the heatsink.
If someone is interested in converting the layout from Autocad to Gerber, he may contact me.
Patrick
PS I did not quite keep the promise of mounting the Loveltechs side by side. But they have such short leads that such a layout would require a separate heatsink for the JFETs.
Zen Mod, I actually wanted to hook up my test speakers just for fun, but didnt get around to it.
Patrick thats a nice compact layout. Nice little board I think that, with all the fets sitting so close, the temp. tracking will be fairly good.
It would be nice if someone could make a B/W pdf of the tracks.
Etching a pair of those, would make for a nice little project
Just keep in mind that the DAO JFET Follower, needs fairly big heatsinks.
Steen
Patrick thats a nice compact layout. Nice little board
I think that, with all the fets sitting so close, the temp. tracking will be fairly good.It would be nice if someone could make a B/W pdf of the tracks.
Etching a pair of those, would make for a nice little project
Just keep in mind that the DAO JFET Follower, needs fairly big heatsinks.
Steen
rjm said:
Richard's 3 channel version also has the advantage that the PS is inconsequential as it cancels out the hum. It sounds quite good, too. I use one as a driver after a 6N6p parafeed linestage (w/ Magnequest B7 iron) and it is about as good a driver for a set of Grados as I've heard.
Here's that thread: http://headwize.com/ubb/showpage.php?fnum=3&tid=6282&fpage=1
Attachments
> It would be nice if someone could make a B/W pdf of the tracks.
You are a lucky guy ....
Quality not brillant, but good enough for DIY.
> Just keep in mind that the DAO JFET Follower, needs fairly big heatsinks.
You have approximately 20~25W to dissipate, so 1°C/W heatsink, or better.
However, if you have headphones with low impedance (say <100 ohm), then you may consider lowering the rail voltages. I would not go much lower than +/-12V though.
Patrick
You are a lucky guy ....
Quality not brillant, but good enough for DIY.
> Just keep in mind that the DAO JFET Follower, needs fairly big heatsinks.
You have approximately 20~25W to dissipate, so 1°C/W heatsink, or better.
However, if you have headphones with low impedance (say <100 ohm), then you may consider lowering the rail voltages. I would not go much lower than +/-12V though.
Patrick
Attachments
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.