luvdunhill Posted January 12, 2011 Author Report Posted January 12, 2011 The two 5K resistors that meet at the point marked FB on the original schematic. I most likely won't do that. I'm happy with the range of adjustment I have, given the heat sinks I'll be using.
luvdunhill Posted January 12, 2011 Author Report Posted January 12, 2011 Since I am only using half of the output devices, any recommendation as to which positions on the board to use? At least I think I only bought that many FETs from you Marc? You can tell how much looking in my parts stash I have (not) done... I'd use the devices close the the output point on the board, then space them out as much as possible along the heat sink. In otherwords, start from the middle and work your way to the edges doing every other position.
Pars Posted January 12, 2011 Report Posted January 12, 2011 The two 5K resistors that meet at the point marked FB on the original schematic. I most likely won't do that. I'm happy with the range of adjustment I have, given the heat sinks I'll be using. Heh. Almost fell out of my chair... R19 and R20. I guessed right. And thanks for reminding me about that schematic as I had forgotten about it. The FB points clear up (somewhat) what is going on with the feedback loop choices. I put stake pins and 2 pos jumpers on for the "switch" for the feedback.
Pars Posted January 15, 2011 Report Posted January 15, 2011 (edited) Marc, From a previous post, this is the current recommendation? My recommendation was R17, R18 = 10K and a 20K pot with a 3.3K series resistor. The two 5K resistors at the feedback point are also changed to 2K. I plan on trying this and going from there. The series resistor is R62. This is not shown in the Eagle files I have from Jacob, which must have been a bit prior. My eagle files also have TO-220 vregs for the servo, etc. on it, which are not used if you are using the OPA445s for the servo opamp. The board itself has positions (unmarked) for TO-92 regs (7815/7915?) on it. I am going to jumper these and go with the OPA445. I am making updates to the BOM as I go. Also, I was missing the 620/680R, was short on 5K and 30K resistors, so when I did my Partsconnexion order, I used Dralorics here (626R, 5.05K and 30K), which are 0.25% or better resistors. Saved me from making a Mouser order, though that will still happen EDIT: anyone have the final Eagle files? EDIT2: For the 10R resistors, could we use 1/4 or 1/2W PRPs? Or do these need to be 1W? (Reason: I have some 1/2W PRPs) Edited January 15, 2011 by Pars
DigiPete Posted January 15, 2011 Report Posted January 15, 2011 Pars, I was going to use half the output devices as well, according to Kevin's original schematic. Not driving K-1000s so that should be fine. Pete
Pars Posted January 16, 2011 Report Posted January 16, 2011 2SA1145 / 2SC2705? Where to buy? I just looked thru my stash and don't have enough of these to do a pair of boards. Any suggestions on where to buy these? I assume Y grade is what I want (not O)? B&D has a few (only 6 SCs), and high price. Mouser has O grade for really cheap. Any other sellers? Anyone on ebay that can be trusted?
Pars Posted January 16, 2011 Report Posted January 16, 2011 2SA1145 / 2SC2705? Where to buy? I just looked thru my stash and don't have enough of these to do a pair of boards. Any suggestions on where to buy these? I assume Y grade is what I want (not O)? B&D has a few (only 6 SCs), and high price. Mouser has O grade for really cheap. Any other sellers? Anyone on ebay that can be trusted? Mouser it is! EDIT2: For the 10R resistors, could we use 1/4 or 1/2W PRPs? Or do these need to be 1W? (Reason: I have some 1/2W PRPs) Looking at the schematic for these (R31-R34), it appears that the CCS's are setup for ~4.5mA, so that is all that is running thru these (assuming I calculated this correctly... 1.6V LED - Vbe ~ 1V / 220 ohms). Footprint is identical to the other leaded/SMD resistor combos, so I am using 1/4W PRPs here.
luvdunhill Posted January 17, 2011 Author Report Posted January 17, 2011 Don't forget that parts often have higher power ratings to deal with failure conditions and control what exactly melts
Pars Posted January 18, 2011 Report Posted January 18, 2011 (edited) R31-R34 don't look like they are set up for any power running thru them: Edited September 2, 2017 by Pars
luvdunhill Posted January 18, 2011 Author Report Posted January 18, 2011 (edited) usually they don't. However, what if one of the CCSes dies (or a transistor), or gets shorted, or what happens as the CCS goes from cold at turn on. I played this game with the T2 to no end. Every single time Kevin's power recommend was spot on, despite me understanding why initially. Just food for thought, assuming this 1W was KG's recommendation, which I haven't verified. It could be a mistake. Edited January 18, 2011 by luvdunhill
Pars Posted January 18, 2011 Report Posted January 18, 2011 Yes, I agree, but I do not know where the 1W recommendation came from. It isn't noted in Kevin's hand-drawn schematic, and in Jacob's (j4cbo) first pass BOM spreadsheet, they are shown as RN55s or 1/4W PRPs.
luvdunhill Posted January 18, 2011 Author Report Posted January 18, 2011 Yes, I agree, but I do not know where the 1W recommendation came from. It isn't noted in Kevin's hand-drawn schematic, and in Jacob's (j4cbo) first pass BOM spreadsheet, they are shown as RN55s or 1/4W PRPs. Sounds like a mistake. The buffers do run warm though, not related to this, but an observation.
Pars Posted January 19, 2011 Report Posted January 19, 2011 (edited) My recommendation was R17' date=' R18 = 10K and a 20K pot with a 3.3K series resistor. The two 5K resistors at the feedback point are also changed to 2K. I plan on trying this and going from there.[/quote'] Marc, I put 20K resistors in (R17-R18), but guess I will switch them to 10K. The pot, VR3... what is this used for and any suggestions for presetting it? Edited January 19, 2011 by Pars
luvdunhill Posted January 19, 2011 Author Report Posted January 19, 2011 That is the output stage bias pot. It need to be set at it's maximum value.
Pars Posted January 19, 2011 Report Posted January 19, 2011 So by maximum value, you mean highest resistance? Should I replace the 20K resistors with 10K? For initial startup, should I reduce the resistance (pot) somewhat until I can monitor it (and increase)? Is the consensus for 1/2 populated output stage current to have a target of 75mA? 150mA? Or something in between?
luvdunhill Posted January 20, 2011 Author Report Posted January 20, 2011 I'd use the 20K resistors. Yes, it needs to be at maximum resistance. Turn it on and monitor the voltage across the output resistors. Let it warm up and turn it up (bias up, trimmer down) until you can hold your hand on the heat sinks for 3 seconds.
Pars Posted January 22, 2011 Report Posted January 22, 2011 Did some reading on headwize, so the max resistance sets the lowest biasing; decreasing this increases the bias current... DOH OK, got my 50 each 'O' A1145/C2705s from Mouser and did some matching last nite (oh fun ). They were pretty consistent but not matching well between NPN/PNPs. The NPNs ran from about 95 to 114 hfe, while the PNPs were from 118 to 155, with all but 6 at 140+. Device classification range is from 80 to 160 for the 'O' devices. I do have some 'Y' devices with hfes in the 150 range for the NPNs (not enough, but some), so I may use this approach for anything deemed critical. Speaking of which, are there any areas where the devices should be matched well? Output FETs obviously, and the 2SK/2SJ parts should be as well. I mocked this up on the schematic indicating areas where matching might be important. Any comments?
Pars Posted January 26, 2011 Report Posted January 26, 2011 (edited) Making progress. One board done other than the output devices, the other almost at this point as well. Fire in the hole this week sometime Will only be 1/2 populated on the outputs. Will probably begin testing with a single pair. Edited March 10, 2018 by Pars updated photobucket crap
luvdunhill Posted January 26, 2011 Author Report Posted January 26, 2011 you're probably better off just installing a single trimmer on CCS and using a fixed resistor on the other. What you have is fine for now, but once you start playing with it, it's easier to adjust one side and then get it close enough that the servo can handle things from there on out. I bet it even sounds a bit better that way.
Pars Posted January 26, 2011 Report Posted January 26, 2011 I like using Amb's adjustment procedure. Used it on the Dynahi and it worked well. http://www.head-fi.org/forum/thread/165788/dynahi-dc-offset-adjustment#post_1960704
mypasswordis Posted January 26, 2011 Report Posted January 26, 2011 Cool looking TO-92 heatsinks! Was it hard to fit the transistors in with the thermal paste?
luvdunhill Posted January 26, 2011 Author Report Posted January 26, 2011 (edited) ok, but remember that the thing adjusts differently than the Dynahi (note amb mentions "bias" and well, you have a separate pot for that). I'd just set them at 500 and leave them and use one to adjust for offset. Like I said, I'd only install a single pot and a fixed 500 ohm resistor. Also, be careful in installing less output mosfets. You'll likely need to change the Vbe multipler values (the pot and series resistor) to accommodate. I'm sure I mentioned that somewhere in this thread, in response to someone wanting exact values for the parts. It will take experimentation, sorry. Edited January 26, 2011 by luvdunhill
Pars Posted January 26, 2011 Report Posted January 26, 2011 The TO92 dual heatsinks came from Marc (from a diyaudio group buy I believe). No, it was pretty easy to insert them with the Arctic Silver... just messy. Marc: do you think I would be better off to just put all the FETs in? As for the "Vbe multiplier" (have to look that up), I just need to ensure that i have adequate adjustment range? Per our conversations, I have 20K R17/18, a 3.3K R62, and a 20K pot currently. Oh, BTW, do you have any extra FETs?
Pars Posted January 26, 2011 Report Posted January 26, 2011 (edited) double post Edited January 26, 2011 by Pars
luvdunhill Posted January 26, 2011 Author Report Posted January 26, 2011 (edited) The TO92 dual heatsinks came from Marc (from a diyaudio group buy I believe). No, it was pretty easy to insert them with the Arctic Silver... just messy. Marc: do you think I would be better off to just put all the FETs in? As for the "Vbe multiplier" (have to look that up), I just need to ensure that i have adequate adjustment range? Per our conversations, I have 20K R17/18, a 3.3K R62, and a 20K pot currently. Oh, BTW, do you have any extra FETs? Here's what I would do. I'd try two pairs, have a DMM across one of the output resistors and turn it on and let that value settle. I'd leave it on less than 30 seconds or so, no longer. And immediately turn it off if say current is insane (1A) (I'd hook it up to a heat sink, but you could perhaps skip this if you're really careful). While it's on, verify that turn the pot up increases this value. Extra points for using two meters across, one for each FET. It might take a bit of travel for it to come up, but when it does it will come quick in this configuration. Record these values. Add another pair of FETs and repeat. Reset the pot. Again, one meter on one of the n-ch and another on the one p-ch output. Record these values. If these numbers are high (high, meaning relative to what you think your heat sinks can provide, that number is up to you to decide.., but post them here and we can philosophize) then we can go from there. I'm 80% sure it will be okay with 4 FETs but doing it with 2 might be a good exercise. I don't have any devices readily available Edited January 26, 2011 by luvdunhill
Recommended Posts
Create an account or sign in to comment
You need to be a member in order to leave a comment
Create an account
Sign up for a new account in our community. It's easy!
Register a new accountSign in
Already have an account? Sign in here.
Sign In Now