[FallenAngel]

Hey guys, so I'm biasing my Dynahi here and I notice a little trend - I bias to 0.75V across output resistors, wait for amp to heat up for 15 minutes, see the bias go up, turn it down, wait another 15 minutes until it goes up again, turn it down again... and so on. It's been running on my desk for over an hour at this rate and I just about maxed out the trimpots. When is "enough"? Should I be expecting that eventually this would not happen? I'm thinking that the amp warmed up enough that the transistors are running at a certain temperature and self-bias but if I turn them down too much, they'll never even reach 0.75V.

EDIT: I have 500R resistors across the 10K trimpots now, but those are being chopped and I'm sticking 1K in there instead for even more adjustment.

 

[digger945]

Replace the 500 with ~620 is what I did.

 

[Pars]

Yep. 500 ohms in parallel with a 10K pot will never get above ~476 ohms. 620 ohm fixed resistor will allow up to 583 ohms.

 

[FallenAngel]

I've got a pair of 680R coming tomorrow but I have 1K around so I'll just use those tonight.

EDIT: Well, 1K in parallel with 10K certainly makes adjustment fun but after a while, I got it to 750mV and pretty stable there, but does still seem to go up ever so slightly over time. After an hour or so, dropped it to 740mV and it's again slowly climbing up, about 1mV every few minute.
EDIT2: Finally looks like it's stabilized at 742mV, not moving much in either direction.

 

[vixr]

FallenAngel...you got PM

 

[rsaavedra]

As most of you might know, the blackout Headfi suffered in 2007 made us lose all photos that had been uploaded onto the Photo gallery, braking all links that pointed to those pics.

I recently uploaded all my Dynahi construction related pics onto Picasa, so here are the links, for anyone who might be interested:

Picasa Web Albums - Raul - Building the Dynahi

Picasa Web Albums - Raul - Finishing the Dynahi

 

[bhjazz]

Good to see it has stabilized for your, FA. The heat of the components, the effectiveness of the heatsinks, the heat of the air inside the case...I'm not surprised it took an hour!
I, too, used 620s for the build I was repairing. It was stable at...about 970 (before the repairs, anyway...)!

 

[FallenAngel]

Yep, and I just got the chassis this weekend so I get to build it. FPE panels won't be arriving until next week, so I get to run it naked for a while. I foolishly ended up ordering only 2 680R resistors instead of 4, used 1K for other position, so one board has one of each.

 

[bhjazz]

Let me know if you need those last two. I'm sure I have extras!

 

[FallenAngel]

Thanks for the offer, but I biased it all with 1K's, just takes a little longer and requires a finer touch.

 

[luvdunhill]

Chris: How complementary did you find the output devices? If you're calculating hfe (rather than grouping by a related parameter, not sure how you did it), could you get identical matches from the npn and pnp device pools?

 

[Pars]

Not sure if you were asking me Marc, and meant to post in this thread instead, but I wasn't able to match the hfe of pnps vs. npns. Apparently it is possible, but the pnps were rather high, at least the ones I got from Digipete. In fact most of them measured out of spec. Y grade is supposed to be from 120-240, but I had hfes going up to 299. I used basically the circuit in post 5 of that thread. I also used a piece of 1" x 1/8" aluminum clamped (clothespin) to the DUT, and was able to run them long enough that the readings were pretty stable, and the device was not getting hot. It was not a problem to hold them after testing. I went thru the setup 3 times and verified all connections, though I was using a plugin breadboard. This is what the last set of measurements looked like:

Code:

[left]measuring Vrbmeasuring Vrb 2SC3421 NPN2SA1358 PNP #HfeIc (mA)VrbRbIbdgpete#HfeIc (mA)VrbRbIbdgpete 2815676.6724.4349692491.60.029212.762.6729.3999731294.70.0 29157.677.524.4349692491.6158.62821563.3629.3999731294.70.0 26160.178.7224.4349692491.6164.316217.564.0629.3899731294.60.0 5162.179.6624.4249692491.40.0321864.0829.3299731294.00.0 2163.680.3624.4149692491.2167.714218.364.2829.3799731294.50.0 21163.680.4224.4349692491.6165.526220.464.9229.3899731294.60.0 3016480.6524.4349692491.6166.76221.265.1429.3799731294.50.0 9164.180.6224.4249692491.4166.418221.865.3429.3899731294.60.0 19164.180.6824.4349692491.60.02223.165.7329.3899731294.60.0 18164.881.0424.4349692491.60.015224.366.0929.3899731294.60.0 25164.981.0524.4349692491.6171.227224.466.1429.3999731294.70.0 17165.281.2124.4349692491.6168.123225.466.3929.3899731294.60.0 7166.781.9424.4249692491.4167.62422766.8629.3899731294.60.0 316782.0524.4149692491.20.017228.667.3429.3899731294.60.0 22168.682.8924.4349692491.60.07232.368.4129.3799731294.5187.4 27168.682.8724.4349692491.6172.231248.273.1529.3999731294.70.0 10169.483.2524.4249692491.4172.85249.273.3329.3599731294.3215.9 11170.383.7324.4349692491.60.030262.577.3529.3999731294.7215.9 2417184.0824.4349692491.6174.710266.678.5229.3799731294.5217.3 12171.184.1224.4349692491.6175.521267.378.7429.3899731294.6218.2 4171.584.2824.4249692491.40.01267.778.7929.3599731294.3225.1 8171.584.3224.4349692491.6168.7926878.9229.3799731294.5220.2 1171.884.3624.449692491.00.020268.379.0529.3899731294.6220.0 16173.285.1724.4349692491.60.019269.979.5329.3999731294.70.0 23173.985.4924.4349692491.60.08271.379.9229.3899731294.6217.8 31174.685.8324.4349692491.6176.322273.480.5529.3899731294.6222.1 14178.787.8624.4349692491.60.032274.680.9429.499731294.8220.0 13179.788.3724.4349692491.60.012277.281.6429.3799731294.5228.1 6179.888.3524.4249692491.40.013279.182.2329.3899731294.6224.3 2018490.5124.4449692491.80.04286.584.3829.3799731294.5241.4 1519696.3824.4349692491.6199.225288.785.0729.3999731294.7236.3 32024.4349692491.60.011299.288.2329.4199731294.9215.9[/left]

 

[luvdunhill]

that's strange they are out of spec... I've never seen any Toshibas that were out of spec like that, with regard to the datasheet. Also, I'm concerned that you could hold the devices after you matched them, that seems strange to me. Sorta surprising that these aren't more complimentary!

Totally off topic, but I learned of a very beautiful thing recently... GR1, BL1, etc. grades. They are even narrower grades that customers paid to basically get matched devices right off the assembly line. When I was talking with Linear Systems for a group buy I did, they offered a similar service... lame thing about Linear is they don't mark the packages when they grade the devices... sorta stupid IMHO. Anyways...

Also, do you need more devices to match with?

 

[Pars]

Yes, I thought it was strange also, which is why I redid my measurements 3 times, with different base resistance. Digipete had measured his PNPs with ~79K Rb, but was just doing them for 15 seconds or less (no heatsink). From watching the Ic as they heat up on the sink, I can see where his measurements would be lower than mine, as initial Ic would typ. be in the ~70-75mA range but would settle in the low to mid 80's. All but 2 of the out of spec devices were the ones I got from him; the others I got from Scott (digger945). I have them colorcoded in my spreadsheet, but that didn't come over when I pasted the measurements in

Sure, I could use more, but I think I have acceptable matches for a 2-ch Dynahi. Scott had posted the close matches he got from Dan Gardner, and they were all in the 153 hfe range or so, so I guess it is possible to find good PNP-NPN matches. Going back thru that thread, it looked like dip16amp (sp?) also had close matches, but used O grade devices for the PNP.

I'd agree with your assessment of Linear... kind of stupid to do tight grading and not mark them

alm:

 

[rsaavedra]

Here's a post for anyone interested in the power consumption of the Dynahi.
 
Purchased a Kill-a-watt meter from Amazon recently, and I've been having some fun measuring a bunch of stuff at home. Among the measured things, my Dynahi, of course.
 
Keep in mind, I built my Dynahi using two Condor power supplies, not Gilmore's designed power supply (in case that might make any difference in consumption.) 
 
Playing Prodigy's "Fat of the Land," and driving two headphones simultaneously: HD600 and HD650. Volume at ~10:30 am, which is the one I commonly use with these headphones. Here's the values I measure: 
 
Input AC: V = 119.5 V, F = 59.9 Hz
 
Measured Power Consumption of my Dynahi:
0.93 Amp
84 Watt / 111 VA
PF = 0.75
 
Right after powerup the amp was drawing only 0.84 amps, and pf was 0.82. But after 25 minutes the measured values were already as shown above, and they remain the same and stable after 40 min.