The parallel o/p in the Rotel 850 doesn't have them, and Pete Maddex's HackerNap doesn't either. I'm aware that being right is not a democratic process.
https://pinkfishmedia.net/forum/thr...output-transistors.216188/page-2#post-3412109
BugBear
Paralleled Output trannies - NCC and HackerNAP
- Thread starter bugbear
- Start date
The parallel o/p in the Rotel 850 doesn't have them, and Pete Maddex's HackerNap doesn't either. I'm aware that being right is not a democratic process.
https://pinkfishmedia.net/forum/thr...output-transistors.216188/page-2#post-3412109
BugBear
I think the Sanken 2SA1216/2sc2922 are even more "internally multiple" than that!
BugBear
I noticed that people use 15030/31s to succesfully drive parallel output devices, but what about MJE243/253s?
And did you use 0R22 or 0R33 resistors on the output devices?
I would like to give it a try: will a bias of 14mV across both resistors be fine? 7mV is standard for an original board (and for close clones), so I thought double is a good starting point, but I am not sure about that...
I think you can only increase the bias significantly if you've moved TR5 to the heat sink so that you get better thermal tracking of the output devices. As far as I'm aware this applies to both quasi complementary and fully complementary output stages. In fairness even if you keep the current/voltage the same in each device the total current will double anyway so you need to check you have adequate heat sinking and moving TR5 is probably a pre-requisite for this.
The parallel o/p in the Rotel 850 doesn't have them, and Pete Maddex's HackerNap doesn't either. I'm aware that being right is not a democratic process.
https://pinkfishmedia.net/forum/thr...output-transistors.216188/page-2#post-3412109
BugBear
I was discussing this with a colleague today who is more knowledgeable than myself and has designed his own class A amps in the past. His conclusion was, as AlanT was saying in another thread, that the extra 100R should be fitted on the Naim/Avondale/Hackernap design...
This is due to the topology of the driver stage and where the 100R is connected to (i.e. across the base emitter junction of the output transistors). In many other amps with parallel o/p devices such as that Rotel and the HoneyBadger on DiYAudio the driver bias is configured differently - feeding directly into the lower driver transistor, which circumvents this issue.
Fitting the extra 100R on the Avondale/Hackernap will ensure the bias current in the parallel output transistors is equal, there will be a small error of a few mA if the extra 100R isn't fitted to the 2nd transistor. Probably not that noticeable in practice which is perhaps why those who haven't fitted them haven't had any obvious issues.
Our analysis also highlighted the importance of making sure the Vbe of the parallel transistors is as close as possible (to a mV or two) if you are also hoping to achieve equal current share in the devices. Hfe matching is also important but perhaps less so than Vbe drop.
S-Man
StrivingON
I reconfigured my NCC300 drivers like this. In theory it enables faster charge suckout from the bases of the output transistors. It also halves the quiescent current in the drivers, which in my case helps with quiescent current stability... but some might prefer more standing current in the drivers - easily fixed by halving the resistor values.
Interesting, thanks for the insight! But what happens to the lower part for quasi complementary?
Last edited:
S-Man
StrivingON
Oops sorry, I am referring to the complementary designs - NCC220 and NCC300.
Leach discusses the triple configurations here:
http://leachlegacy.ece.gatech.edu/lowtim/output.html
The same principles apply.
I'm using the dual version of fig 1 (without a speedup cap).
Leach discusses the triple configurations here:
http://leachlegacy.ece.gatech.edu/lowtim/output.html
The same principles apply.
I'm using the dual version of fig 1 (without a speedup cap).
Interesting, as I was wondering about doing this myself and if there was any disadvantage in the increased quiescent current resulting from the extra 100R method.
I also notice the Rotel topology still connects the feed back point to the midpoint between the driver bias resistors (more like Avondale/Naim) but Honeybadger doesn't connect the feedback point here at all - another subtlety I'm not sure about the consequences of?!
S-Man
StrivingON
I built a double decker NCC300 and the lower deck had less than ideal cooling for the drivers. This tends to increase the quiescent current because the Vbe multiplier is thermally coupled only to the output devices. Halving the current in the drivers halves the dissipation and means the Iq doesn't ramp up as much as the amp warms up. Ideally I would have moved the drivers to the main heatsink but this was impractical.
See Fig 13 here:
http://www.douglas-self.com/ampins/dipa/dipa.htm
I am talking about changing from A to B.
Selfie mentions the theoretical advantages = faster switchoff.
Leach reckons that the speedup cap is not required. I don't use one.
In a different amp, some years back, I did evaluate the subjective effect of putting a cap here - it seems to reduce a bit of treble "sparkle".
a.palfreyman
pfm Member
Cripes. I wish I understood 1/4 of that^^^
Grog
pfm Member
Today I'm finishing up the boards for my HN with paralelled op devices.
I'm strongly considering mounting TR5 to the back of the board so it will be almost or even touching the heat spreader / heatsink.
I'm mounting on a 6mm thick AL plate running the length of the board, this is bolted to the case sides
Is this sensible and any downsides?
Well, with the wet weekend I finally got round to finish putting together my new dual mono NCC220 amp with the hooks in place to go parallel output. I already have an Avondale S100 amp with NCC220 Qudos boards but decided it would be good to start afresh with a new case for adequate heatsinking, plus larger 400VA Canterbury windings toroids, feeding HackerCaps with Kendeils. I also took the precaution of an Avondale SSM2 soft start module and a pair of Dada speaker protection boards.
Stage 1 was just to get it up and running but with the output trannies already selected as matched pairs, but with the lowers not yet connected. Aim is to run it up for a couple of weeks to "burn in" and get accustomed to the sound, then hook up the second trannies.
Longer term I plan to experiment with local decoupling caps across the outputs, as per the Hackernap, and also possibly regulate the front ends with a pair of Avondale VBEs I've yet to build up.
I did my final commissioning today and was hoping to at least get similar performance as the S100 as a baseline. I wasn't prepared for the significant improvement I've already got from the getgo! Whether its the higher spec Trafo's or fresh Caps in the Hackernaps I can't say, but this amp sounds much more open and authorative than my S100, darker background with more detail and bass definition and texture to die for... To say I'm pleased would be an understatement and I haven't even gone parallel output yet!
I'll hopefully work out how to post some pics...
https://www.dropbox.com/s/8jn0pl62unx9jww/DSC_4048_0000100002.jpg?dl=0
https://www.dropbox.com/s/klb9obgfdo6f6ax/DSC_4049_0000100002.jpg?dl=0
https://www.dropbox.com/s/563mdusgggzh735/20211003_172335.jpg?dl=0
https://www.dropbox.com/s/es0o6v068t9xmaj/20211003_172540.jpg?dl=0
Stage 1 was just to get it up and running but with the output trannies already selected as matched pairs, but with the lowers not yet connected. Aim is to run it up for a couple of weeks to "burn in" and get accustomed to the sound, then hook up the second trannies.
Longer term I plan to experiment with local decoupling caps across the outputs, as per the Hackernap, and also possibly regulate the front ends with a pair of Avondale VBEs I've yet to build up.
I did my final commissioning today and was hoping to at least get similar performance as the S100 as a baseline. I wasn't prepared for the significant improvement I've already got from the getgo! Whether its the higher spec Trafo's or fresh Caps in the Hackernaps I can't say, but this amp sounds much more open and authorative than my S100, darker background with more detail and bass definition and texture to die for... To say I'm pleased would be an understatement and I haven't even gone parallel output yet!
I'll hopefully work out how to post some pics...
https://www.dropbox.com/s/8jn0pl62unx9jww/DSC_4048_0000100002.jpg?dl=0
https://www.dropbox.com/s/klb9obgfdo6f6ax/DSC_4049_0000100002.jpg?dl=0
https://www.dropbox.com/s/563mdusgggzh735/20211003_172335.jpg?dl=0
https://www.dropbox.com/s/es0o6v068t9xmaj/20211003_172540.jpg?dl=0
Mynamemynaim
38yrs a Naim owner
Is that an old S100 with replacement 220 boards? I though the sale of S100's predated the 220 by quite a while.
EDIT; does the S100 cap-board have resistors or inductors on it?
HackerCap - same question?
BugBear (just nosy)
Last edited:
Yes the S100 dates back to 2009, but I fitted 220 boards to it a couple of years ago.
The minicap in the S100 has inductors as does the HackerCap - I used 5.6uH 8A in my build which I believe may be what Avondale used. My plan was to try and "clone" the S100 as closely as possible electronics wise so it would be easier to determine what difference, if any, the other components and modifications make.
With that said when I ordered the 6800uF Kendeils from Italy I specified 63V rating, but I received 100V parts! They are noticeably taller than the caps in the S100 and am led to believe may have a slightly lower ESR as a results of the higher voltage rating. It's also worth bearing in mind the S100 has been switched on more or less 24/7 since I've owned it, so the caps in it have had 12 years of solid use so maybe not at their best anymore!
Grog
pfm Member
I'm now at the point of running in my HN with parallel MJL3281's,
in terms of bias, it does shift a lot with temp, probably doesn't help that where I'm testing is about 16~17 degrees ambient at the moment.
aiming for around 3.5mv over each of the 0R33 which is total 21ma. does that seem about right?
in terms of bias, it does shift a lot with temp, probably doesn't help that where I'm testing is about 16~17 degrees ambient at the moment.
aiming for around 3.5mv over each of the 0R33 which is total 21ma. does that seem about right?
Pete MB&D
Pete Maddex, the one and only!
Sounds about right to me, I can't remember what I set mine to, I guess I should check it out at some point.
Pete
Grog
pfm Member
thanks Pete,
if it's of any interest, i used a variable heat gun on a very low setting, and go the case side to ~32 degrees to simulate summer!
the bias rose to about 30ma, compared to cold ~18 degrees it was closer to 15mA.
I'm wondering if some of the differences heard with parallel devices is the bias per output transistor is halved. as well as passing half the signal current.
in theory should we keep the bias current the same per device, i.e. double total bias when doubling up OP devices? to keep the working point of each device the same as single OP devices.
if it's of any interest, i used a variable heat gun on a very low setting, and go the case side to ~32 degrees to simulate summer!
the bias rose to about 30ma, compared to cold ~18 degrees it was closer to 15mA.
I'm wondering if some of the differences heard with parallel devices is the bias per output transistor is halved. as well as passing half the signal current.
in theory should we keep the bias current the same per device, i.e. double total bias when doubling up OP devices? to keep the working point of each device the same as single OP devices.