| |||||||||||||||||||||||||||
| Moderators: cmoy | Welcome. Please log in. | ||||||||||||||||||||||||||
Registration is required to post a new topic or a reply. 
| |||||||||||||||||||||||||||
| tb Member Joined: Jan. 21, 2007 |  Message [#1] posted on: 11-28-2008 01:10 PM CST (US).   |  |  Hello, After reading comment of PPL about quality of the sound of this AD/Elantec combo I got the parts, but never had chance to build it. Finally, I am ready to build the amp., but I can not find schematic anywhere in the posts and project libarary. Can PPL or anyone else publish the schematic? I would like to use it for both headphone/preamp duties. Thank you very much for the help. [Edited by tb on 11-28-2008 at 09:59 PM.] |
tangent  Headphone Council Joined: Aug. 27, 2001 | Message [#2] posted on: 11-29-2008 04:17 PM CST (US). | | If it were me, I'd start with the PIMETA schematic, take out the ground channel, and use a dual-rail power supply. 3-channel amps don't work well as preamps. |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#3] posted on: 11-29-2008 08:28 PM CST (US). | | the el2009 sounds like a fancinating device, on paper at least - I have never heard one. its current capability is like a small power amp! topology-wise, it is absolutely symmetrical, positive half vs. negative half. and each half is essentially a buf634 buffer by a class A emitter follower. having said that, I am not sure what you gain by going with so much current power and so high of a slew rate - they may create more problems for you than issues solved. schematic will be quite similar to any of those composite amps where buf634 is used as a buffer. Walter Jung in his book has a few such examples. and you can find them at his website. |
aos HeadWizer Joined: Jun. 15, 2008 | Message [#4] posted on: 11-29-2008 08:41 PM CST (US). | | I think I have schematics somewhere but I am fairly sure I got it from ppl in confidence so I can't publish it. Try to contact ppl directly. EL2008/9 were legendary. They were designed to drive a bunch of coax cables - e.g. put on a cable TV splitter in an apartment complex to drive a dozen cables at the same time. Wide bandwith, able to drive tough loads at high power. You should be able to use it as a normal buffer in a loop with another opamp, as others already mentioned. |
| tb Member Joined: Jan. 21, 2007 | Message [#5] posted on: 11-30-2008 07:46 PM CST (US). | | Thank you very much for answers. I will look for other schematics. |
| tb Member Joined: Jan. 21, 2007 | Message [#6] posted on: 12-02-2008 09:20 AM CST (US). | | Hello, I did try to contact ppl directly with no avail. Among threads people sometimes use buffered cmoy phrase, but search for more information does not bring any results. So, I have a two questions: can this schematic be used for this purpose with no changes: http://web.telec.../buffer_en.html Pimeta amp is using two transistors Q1 and Q2. Is use of them mandatory? Also, what is the X symbol just after volume control bypassed by C1? |
| tangent Headphone Council Joined: Aug. 27, 2001 | Message [#7] posted on: 12-02-2008 10:05 AM CST (US). | | Well, keep in mind, a "CMoy" is just short-hand on these forums for a fairly generic amp design. I suppose we call it that for lack of a better term. You could call what you're looking for a buffered noninverting op-amp gain stage, but that's not going to be a very helpful search term, either. I'd add a volume control and ditch the 50 ohm output resistor. And naturally, the current adjust resistor they show for the BUF634 has no place on an EL2009. You'd be losing out on class A biasing of the op-amp and the Jung topology, compared to the PIMETA. Of course not, but then, I don't see why you wouldn't if you're going to use up such a special buffer. I'm not a big advocate of trying to build the "final" version of any DIY project from the start...I think there's a lot of value in experimentation and trial tests. But when the chips involved are rare or expensive -- both in this case -- I think there's merit to pulling out all the stops and going with the best well-tested design you can find. I'm pointing you to the PIMETA not because I helped make it, but because ppl helped make it, too. I think if ppl had responded to you, he'd have recommended a circuit like this. On the PIMETA board, C1 is shorted out by a trace, because my default advice is to skip C1. This saves you from having to add a jumper here, at the expense of making things a little harder on those who do want a C1. The X is a hint that this jumper exists on the board, and that it can be cut to put C1 in the circuit. For the details on how to choose, see my article Input Capacitors for Headphone Amps. [Edited by tangent on 12-02-2008 at 10:09 AM.] |
| tb Member Joined: Jan. 21, 2007 | Message [#8] posted on: 12-02-2008 10:45 AM CST (US). | | Hello Tangent, thanks for the prompt response. As you can see from my posts I am a beginner in this field. Using less elements, if it is possible, is welcome. Also, buying parts is a pain. There are no stores, but RS, around my place. I do not have those transisitors. So, I would need to get them from internet stores what takes time. I have also noticed another project, looking simple, Apheared 47. Can this schematic be used for my purpose? |
| tb Member Joined: Jan. 21, 2007 | Message [#9] posted on: 12-02-2008 11:18 AM CST (US). | | Tangent, I have read your response again. I will build by your recommendation Pimeta amp. I will try to find transistors. Can those sold by RS be used in the circuit? Also, Pimeta schematic shows resistor R11 connected to ? what would be a pin #. Would it much EL2009 pin? Thanks again. |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#10] posted on: 12-02-2008 11:50 AM CST (US). | |
yes, it should work. see the attached. I would use a dual opamp to save on real estate. R4's value depends on your gain requirement. my experience is that 3x is about right but you can go higher if your power supply allows or your headphone is quite insensitive. the pin out is per el2009c datasheet, looking at the mark with pin 1 on the left and pin 5 on the right. I add a 110ohm resistor in between the buffer / opamp to just prevent the opamp going haywire should something weird things happen to the buffer. I would absolutely keep R10 in the circuit and outside of the feedback loop: when you insert the headphone plug, you inevitably short the output and you want R10 there to limit the current and protect the buffer. it can be as high as 120ohm but I typically use 47ohm. R10 also lowers the dampening factor and some people argue higher dampening factors of tube amps are what make them sound good - I don't personally buy that theory. I know Pavel and he is very knowledge on anything audio so I wouldn't hesitate take the design, which is the classic composite amp idea that has been used many times over in audio and elsewhere. hope it helpls. Attachment: C4437.gif |
| tangent Headphone Council Joined: Aug. 27, 2001 | Message [#11] posted on: 12-02-2008 01:13 PM CST (US). | | You and almost everyone else in the world. Unless you live in Tokyo or Silicon Valley, you're going to be doing a lot of mail order. No. It's designed to substitute a second op-amp channel for a buffer. I don't like this circuit at all. Give me a real buffer any day. Again, the PIMETA needs mods if you're going to use it as a preamp. You can either replace the virtual ground with a simpler one that doesn't separate input ground from output ground, or you can use a dual supply. Or, you can give up on the preamp idea and go with a pure headphone amp, of course. 3-channel amps are for heapdhones only. As far as I know, RS only sells bipolars, not JFETs. EDIT: You can make some very fine constant current sources from bipolars, but they're not at all like the JFET based one in the PIMETA. One I particularly like is in the YJPS schematic: D2, R5, R6 and Q1. Great performance, reasonable part count, and current settable as precise as you want by changing one resistor. The only problem is that it's more wasteful than the JFET based one: you lose current down the LED path, and I suspect it won't run down to as low a voltage. Just as in the schematic you linked to above...it's a bandwidth control resistor, a special feature of the BUF634. it doesn't apply to any other buffer. [Edited by tangent on 12-02-2008 at 01:18 PM.] |
| tb Member Joined: Jan. 21, 2007 | Message [#12] posted on: 12-02-2008 01:49 PM CST (US). | | Tangent, I will tweak it as you suggested in the first response, no ground channel. I would like to have an option to use it as a preamp too. It is a noobie question, but can we use darlington transistor for current source purpose? millwood, |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#13] posted on: 12-02-2008 02:38 PM CST (US). | |
it depends on the design. bjts have extremely low saturation voltage so in theory they will hold their current constant longer than jfets or mosfets. the particular jfet CCS is poorly designed in the Pimeta, with an addition of one more resistor, you can improve its performance substantially, making it more like the CCS in PPA. here is a comparison of three CCS, using a bjt/led, a Pimeta-styled jfets and an improved ppa-styled jfets. the bjt CCS is widely used and proven to be very effective. R22 sets the current going through the LED (simulated with three diodes -> 2.1v. most LEDs are in the 1.7v - 2.2v range). the current going through the LED is, in this case, (15-2.1)/2k=6.5ma. typically in the 5-15ma range. R24 sets the current going through the jfet, and R19: =(2.1v-0.7v)/220=6ma. so R22, LED, R19 and R24 will replace R10, Q2 and Q1 in Pimeta, or Q1, Q2 and R9 in PPA. the bottom of the chart is a DC sweep analysis. As you can see, the bjt CCS can work with the voltage drop across down to 7.5v, the PPA-style jfet ccs down to 8v, and the Pimeta-style jfet ccs needs 21v to work - so it is essentially disconnected from the circuit at anything less than +/-21v. you will also notice that when they are working, the bjt ccs and the ppa-style ccs hold their current almost constant - a flat line, while the Pimeta-style CCS's current output varied by 2ma. the point? a bjt ccs is the best of the three in maintaining its current over the widest voltage swings. and it is cheap. Attachment: C4438.PNG |
| tangent Headphone Council Joined: Aug. 27, 2001 | Message [#14] posted on: 12-03-2008 04:33 AM CST (US). | | Sure, but all it is is one bipolar driving another, giving higher current gain. (Plus maybe some protection components.) You don't need the extra gain here, so I'm not seeing the value. Not when you consider all the constraints, one of which is that the board can't get bigger. (It's designed for a particular enclosure.) I could buy the space back by removing R11 or requiring surface mount BUF634s, but both seem like poor compromises to get a better CCS. There's a chance the next version of the PIMETA will use a different buffer, at which time we might have the space for the extra R. Yes, but can we talk ourselves into the extra 3 mA or so you need to run an LED per channel, in a portable design? In the YJPS, no problem, we've got power to burn. In the PIMETA? Maybe not. [Edited by tangent on 12-03-2008 at 04:35 AM.] |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#15] posted on: 12-03-2008 05:37 AM CST (US). | |
if you care about space, you can get rid of the lower jfet and replace it with a resistor - it will hold current steady better than the two jfet version.
if current consumption is an issue, get rid of the CCS all together: I would be surprised if anyone can hear much of a difference in a blind listening test with or with the CCS. |
| tb Member Joined: Jan. 21, 2007 | Message [#16] posted on: 12-03-2008 10:06 AM CST (US). | | Hi, I forgot to mention this amp will not be used as a portable unit. I want to build regular psu and , as an option, battery one. I am planning to use it as a pre. for Gainclone. |
| aos HeadWizer Joined: Jun. 15, 2008 | Message [#17] posted on: 12-03-2008 12:48 PM CST (US). | | That's why I use SST5xx for current sources - no compromise in performance, very small package, works at low voltages. It is tiny - SOT-23 - but it has only 3 pins so it's fairly easy to solder. I don't know if I can hear the difference but I have measured improved performance, in one case 10dB improvement in 2nd harmonic distortion by biasing the amp with a current source. |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#18] posted on: 12-03-2008 12:59 PM CST (US). | |
if it is powered by a/c (eventually), it doesn't matter if you save a few ma of current. if you are using this as a preamp to drive the gainclone, you are unlikely to need so much current capabilities out of the el2009. most opamps will do as the gainclone is unlikely to have a very low input impedance.
TI has some sot23-5/6 opamps that can output over 100ma - great for an ultrasmall headphone amp. |
| millwood HeadWizer Joined: Oct. 30, 2008 | Message [#19] posted on: 12-03-2008 01:46 PM CST (US). | | I did more analysis on the various jfet ccs. we know from the dc sweep that the bjt and cascoded jfet ccs are quite good at holding its current while under wide voltage swings, and the pimeta-styled jfet ccs did the worst. that analysis doesn't say anything about the ac performance of various jfets, especially their stability: as the ccs is being used as a load to opamps. so the following is an ac analysis of the three jfets, with the addition of a simple one-jfet ccs. you will notice a few things: a) all ccs are unstable at high enough frequencies: the curve begin to go up, suggesting their ability to hold constant current weakens at those frequencies. this is quite bad because the CCSs are being dampened by a 1k resistor in our analysis. so if you lower the value of that resistor, the situation gets worse. b) the bjt ccs did the best in terms of ac performance: it saw the least performance deterioriation, and again the pimeta-styled ccs saw the worst performance deterioriation. c) if you look at the flat portion of the curves, you will notice that the bjt and cascoded jfet ccs are the best performers, followed by the one-jfet ccs and the pimeta-styled ccs has the worst performance - consistent with our dc sweep analysis. the point? 1) be careful using a ccs as a load to a high-speed opamp. hope it helps. Attachment: C4439.PNG |
| tangent Headphone Council Joined: Aug. 27, 2001 | Message [#20] posted on: 12-03-2008 06:45 PM CST (US). | | Hmmm. |
| tangent Headphone Council Joined: Aug. 27, 2001 | Message [#21] posted on: 12-03-2008 07:48 PM CST (US). | | I wonder, in your PIMETA-like CCS, did you set the Idss of the lower JFET to be lower than the upper one? The two JFETs aren't supposed to be the same, or if you do use the same part, you have to hand-measure them so the lower one has a lower Idss. |
| Page 1 2 | Next Page | Prev Topic | Next Topic | ||||
 (remove _nospam_ ) |
© Chu Moy, 2001.