| |||||||||||||||||||||||||||
| Moderators: cmoy | Welcome. Please log in. | ||||||||||||||||||||||||||
Registration is required to post a new topic or a reply. 
| |||||||||||||||||||||||||||
dBel84 HeadWize Fanatic Joined: May 5, 2006 |  Message [#140] posted on: 12-02-2006 11:26 AM CST (US).   |  |  That would be interesting indeed Steinchen. This would apply to all CFP designs which I can see me using again. ( and DaveM - sorry for not referencing you, I remembered that it was you after I had posted)..dB |
| Dave_M HeadWizer Joined: Jul. 1, 2006 | Message [#141] posted on: 12-02-2006 01:25 PM CST (US). | | BD139/140 may very well be better for this amp. I'll look forward to the shootout 
You could increase the value of the 0.5 Ohm resistors to something higher like 10 ohms or more. That might help a bit but it's not a very good solution. How about some current limiting in the PSU? 500mA per rail should be plenty. Something similar to the ε12 but for overload protection might not be fast enough to save the output transistors so it should be part of the PSU. [Edited by Dave_M on 12-02-2006 at 01:32 PM.] |
amb Headphone Council Joined: Apr. 1, 2004 | Message [#142] posted on: 12-02-2006 03:36 PM CST (US). | |
It only makes sense if you have a balanced source, and then you'd need two boards. Since the board is designed to fit neatly into a Hammond 1455Q case, and there isn't such a case long enough to fit two boards and still have enough leftover space for a 4-gang pot and input/output jacks, you'd have to use some other casing solution.
The output resistors' value cannot be increased without also affecting the bias resistors R18/R42. Also, since there is no global NFB to lower the output impedance, increasing the output resistors' value will also raise the output impedance directly. Under no-signal conditions, there is no problem. The output should be very close to 0V because the DC servo keeps it there. Assuming 10mV DC offset, the 0.5Ω output resistors will limit the maximum short circuit current to 20mA, which is plenty safe. A very short overcurrent spike from the discharge of the 220uF rails capacitors would occur if the output short is during maximum output. After that the 7815/7915 voltage regulators will current limit above 1.5A which is right at the limit of the BD139/140. But this is an extreme condition. People should develop the habit of turning down the volume before inserting/removing the headphone plug. [Edited by amb on 12-02-2006 at 03:39 PM.] |
| Clutz HeadWize Fanatic Joined: Jun. 1, 2002 | Message [#143] posted on: 12-02-2006 03:57 PM CST (US). | | I'm curious- Why use BJT outputs instead of MOSFETs? |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#144] posted on: 12-02-2006 05:19 PM CST (US). | |
MOSFETs would require much higher idle currents to work well, which would necessitate heatsinking, and make larger demands on the power supply. The amp would have to be bigger as a result. |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#145] posted on: 12-07-2006 01:46 AM CST (US). | | An update: - I have secured a number of 2SK170BL and 2SJ74BL JFETs which are on their way to me. I'll sort through them to see how well I could get matched pairs out of them, and will offer them as options along with the board. |
| awpagan HeadWizer Joined: Jan. 6, 2005 | Message [#146] posted on: 12-07-2006 03:09 AM CST (US). | |
How close do these fet's\tranny's need matching? allan ps would be easier matching pnp to pnp and npn to npn |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#147] posted on: 12-07-2006 04:21 AM CST (US). | |
I think within 10% is good enough, if you could get better matches, more power to you.
Indeed. On this amp, due to the complementary topology, if one is to match the transistors at all, it should be between the N and the P devices. My past experience with the 2SK170BL and 2SJ74BL have been reasonably good, but I'll have to see if this new batch is as well behaved as before.
True, but it's not meaningful to do that for this amp. |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#148] posted on: 12-07-2006 04:30 AM CST (US). | |
I think within 10% is good enough, if you could get better matches, more power to you.
Indeed. On this amp, due to the complementary topology, if one is to match the transistors at all, it should be between the N and the P devices. My past experience with the 2SK170BL and 2SJ74BL have been reasonably good, but I'll have to see if this new batch is as well behaved as before.
True, but it's not meaningful to do that for this amp. P.S. runeight has convinced me to change the name of this amp to "Cavalli-Kan Kumisa III" (CKČIII). This name will appear on the production board. However, for forum purposes we could just use "CKČ", "CK^2" or just "CK2". [Edited by amb on 12-07-2006 at 04:30 AM.] |
| Dave_M HeadWizer Joined: Jul. 1, 2006 | Message [#149] posted on: 12-09-2006 07:35 AM CST (US). | | Why do the JFETs need to be closely matched? What would happen if they were very poorly matched? |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#150] posted on: 12-09-2006 04:17 PM CST (US). | |
Mismatched transistors cause two main problems in a complementary topology like this amp: 1. Higher DC offset 2. Higher distortion The DC offset issue is dealt with by the DC servo, but it will take some amount settling time. At turn on, the better matched the transistors, the less turn on transient there would be. Since this amp has no global negative feedback for distortion reduction, the second item is of more interest. Not only would the JFETs benefit from matching, the rest of the BJTs would too. The problem is that these JFETs are a little harder to come by than the other transistors, and JFETs tend to exhibit more Idss variation from device to device than a BJT's Hfe. It would take a larger sample of devices to achieve good matching, and this is why I am going out of my way to offer pre-matched JFETs for everyone. It's gonna be a lot of work! [Edited by amb on 12-09-2006 at 04:18 PM.] |
| awpagan HeadWizer Joined: Jan. 6, 2005 | Message [#151] posted on: 12-09-2006 05:08 PM CST (US). | | how about complementary fet's on the same die? found a variety but mainly surface mount 8pin packages(ssop i think) allan [Edited by awpagan on 12-09-2006 at 05:12 PM.] |
| awpagan HeadWizer Joined: Jan. 6, 2005 | Message [#152] posted on: 12-09-2006 05:16 PM CST (US). | |
|
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#153] posted on: 12-09-2006 07:32 PM CST (US). | | The 2SK170/2SJ74 JFETs do have some nice properties that make them good for this use, such as low noise and high Gm. Most of the "complementary pair in a chip" devices, aside from being MOSFETs instead of JFETs, are intended more for switching apps. At any rate, we're committed to the Toshiba devices now (and may consider the Linear Systems LSK170/LSJ74 after they release the P-channel version). |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#154] posted on: 12-11-2006 05:49 AM CST (US). | | Hi folks, thanks for your interest. I will be doing one final check of the layout, and order the production boards this coming weekend.  I will post relevant updates when there is new information. |
twofishy Member Joined: Nov. 14, 2006 | Message [#155] posted on: 12-13-2006 08:34 AM CST (US). | | (moved post to marketplace) [Edited by twofishy on 12-13-2006 at 08:47 AM.] |
coffin1967 Member Joined: Nov. 4, 2006 | Message [#156] posted on: 12-17-2006 09:08 AM CST (US). | | Hi Is all the bjt can be replaced? what is the criterion? Cause I got some Japanses BJTs on hand, such as 2SC2240/A970, C3421/A1358 cheers Coffin |
runeight Headphone Council Joined: Mar. 8, 2002 | Message [#157] posted on: 12-17-2006 02:56 PM CST (US). | | Hello Coffin. The bias points in the amp are set for the BC550/560. If the transistors you are using have similar characteristics you migt be able to use them. But if they are too different from the BC series, then the bias points will be off and the amp won't behave as well. It might not even work, depending on the differences. The BC550/560 and BD139/140 seem to be generally available worldwide and I would recommend that you find these if possible. |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#158] posted on: 12-17-2006 04:54 PM CST (US). | | Production boards have been ordered. See my marketplace post for details. The production board is v6.08, with only a couple of small adjustments since v6.07. The C37/C38 capacitors are now 10.5mm diameter 5mm pitch, and C39/C40 are still 5mm pitch, but their footprints have been widened to accept fatter film caps. I had to move some other parts slightly to make room for this change. The following is the new layout and 3D rendering. I will post a preliminary BOM very soon.
Attachment: C2392.png,C2393.jpg |
tomb Headphone Council Joined: Mar. 1, 2006 | Message [#159] posted on: 12-17-2006 05:52 PM CST (US). | | Amb/Runeight, With an offboard power supply, would it be feasible to put the board in a 1455N1201? That would put the board sideways, meaning the the output trannies would be against the slots on one side. Is the heat transfer sufficient with the trannies against the slots? If not, would heat sinks bolted through on the outside make up for the lack of contact area? As mentioned earlier, I'm planning on putting the Amveco in a walwart case and running two sets of lines/plugs. So, I'm thinking I would not need the extra room offered by the 1455Q. EDIT: Scratch that - 4.2" is too big for the Hammond's 4.06". For some reason, I thought it was closer than that. For future reference, an answer to the general question of output trannies against slots would be appreciated, though. [Edited by tomb on 12-17-2006 at 05:59 PM.] |
| amb Headphone Council Joined: Apr. 1, 2004 | Message [#160] posted on: 12-17-2006 07:55 PM CST (US). | |
Hi tomb, if you bias the output transistors to 30mA, then you don't need heatsinks. Each output transistor will dissipate 0.45W which will make them pretty warm, but should still be ok without heatsinks. If you set the idle current any higher, though, then they should be heatsinked. As for how big a heatsink to use, it depends on how high you set the bias. For this amp I don't see a need to go much above 30mA, though. The slots on the side of the Hammond cases do not provide a good heatsink surface. So if you want to heatsink the transistors, use real heatsinks or use a case where you could bolt the transistors to a flat case wall (with appropriate TO-126/TO-225/TO-220 insulators, of course). |
| Page 1 2 3 4 5 6 7 8 9 10 11 12 13 14 > >> | Next Page | Prev Topic | Next Topic | ||||
 (remove _nospam_ ) |
© Chu Moy, 2001.