I'm never 100% on these things, but here's some suggestions re: a path I might go down if I were in a similar situation and wondering about the caps. I also type a lot... and quickly... and more conversationally... so... I'll likely make some errors.
re: C1 and C2. I'd look up "AC Coupling".
Note also, that only one of the bias supplies is referenced to ground, but caps are added to both.
I think I now understand what you meant earlier re: "a steady/constant source of DC injected into the AC signal", when discussing the bias voltage. I may also have unintentionally misled you with my reply. You are correct that the AC signal would now have a "DC offset" relative to some arbitrary point. It doesn't "alter" the signal (or it shouldn't). The signal just 'rides along'. A cool experiment might be to scope the signal and toggle the AC coupling. The signal's amplitude and frequency etc. should be unaltered, but it will have a DC offset relative to the 'ground'. I'm lousy at creating plots, but I'm sure there are some cool ones already out on the interwebs somewhere. I put mine below.
The intention (here and in most Class A amps) is to make sure that no matter what the signal's voltage (amplitude) that when it's riding along on that DC... that the difference in potential between the gate and source is always high enough to where the devices don't turn off (loosely speaking... in push pull amps there are differing ideas not relevant to the discussion). An interesting (to me) and easier to illustrate example might be to assume a device will turn on when the Vgs is 1V => VGS_t = 1V. Any less than that, and it's "off". Let's assume no bias supply first. Each end of the secondary hooked to the gate and source. The orange line is the potential difference between the gate and source through time. Let's assume the signal through the secondary has a peak to peak of 2V. Here's my attempt... prepare for the gallery to laugh with glee (or disgust). Then, here's the signal with no DC offset and the signal with 1 and 2 volts of DC offset from our bias supply. When would the device be "on and off" for each case? Black line for a hint. Again, I may have this COMPLETELY wrong, but it's they way I've thought about it.
Now back to caps. We now understand why we need a bias supply. So, why are the caps in the circuit once Q1's bias voltage is referenced to ground vs. the source pin of Q1?
- For Q1's bias supply, imagine perhaps that ground is connected to one speaker terminal and out is connected to the other... picture what might happen to your speaker's voice coil with the cap... without.
- Might any instabilities and/or settings on the amp have the potential to create even a temporary DC current with a meaningful enough voltage potential and route it through my speakers? (Nelson hints that referencing the supply here solves one problem, but he's also moved along in the discussion to making it more of a "realistic" design vs. a concept with ideal parts.
- Will this circuit amplify DC?
- Do any of my sources have the potential to output DC or enough DC to matter if it's amplified and reaches my speakers?
- Am I going to use any type of speaker protection circuit?
- Will anything you're proposing re: modifications to the circuit alleviate the concerns re: potential DC reaching my speakers?
- Am I simply trying to reduce parts count and/or do I 'not like' coupling caps?
Random thoughts before my second cup of coffee. Might be gibberish, but maybe a piece of it is helpful. I had some fun thinking it through and typing it up... 🤷♂️
Last edited:
Excellent first cup theory, @ItsAllInMyHead!
You really do a nice job retracing my impressions of the lecture. A few of my errors you point out I too have caught during my second round of study and I am in total agreement. A big "aha" I have not considered is your point about the bias being in stream with the speaker. Your rational makes a lot of sense. I also had reached a tentative conclusion that the caps are necessary but more from the standpoint that I wasn't quite tracking NP's progression of the circuit from simple to less simple.
"Will anything you're proposing re: modifications to the circuit alleviate the concerns re: potential DC reaching my speakers?"
Yes, IMHO. I'm still brought back to NP's comment about powering the bias V with a battery. This implies to me that you don't have to tie into the output, i.e. the source pins. It seems that he is powering the bias voltage using the source pins out of necessity, not out of "audio-motive" requirements. Going on logic regarding what bias V is on paper, a low voltage source, I see no reason to tie externally supplied DC to the source pin, as it would be redundant. That leaves only a few options for how to supply the external DC; the option that jumps out is tying in between the gate and ground. That's my current guess on how to do this. I like this setup because you could have a standalone bias supply, adjustable, that you can plug in to various builds like you would a mains supply. So whenever you tinker on new circuits that is one less variable to contend with.
I have made some progress elsewhere on this general concept. Firstly, It's looking like my idea to use Arnon7 to roll my own transformer is not going to cut it. That particular core material is very efficient based on being very thin laminations, .007", but it doesn't quite hold up at high frequency. It's good but not as good as the nickel alloy used in the Jensen. So that little guy is going on the backburner.
Secondly, I have gotten pretty far into the design stage for a heatsink die to use with the dual channel IC. It looks pretty doable for its tiny pad size. They way it works is each pad gets it's own copper billet that expands outward from the pad footprint. The large billet pads provide enough workspace to implement various electrical connections. The copper billet is sunk to aluminum nitride insulation which is then sunk to an aluminum housing that will see direct contact with a liquid cooling loop. Because I don't know if using a dual channel IC is unwise, I made provisions to assimilate standard through-hole ICs of the single channel type. Basically it's a modular breakout for power transistors. After reading through various threads on FETs one more time, the IC I mentioned earlier still looks very appealing. The only drawback I can imagine is if there is a proximity issue that affects sound quality due to isolation problems. Isolation aside, I would think they should be very closely matched; just speculation. At this point the discussion is afield of the build guide so any more on it should probably get moved to it's own thing. I'll build the die first and if that goes well I start a thread about working through the concept and figuring out what works and what doesn't. Here's the die concept in both IC configs.-
You really do a nice job retracing my impressions of the lecture. A few of my errors you point out I too have caught during my second round of study and I am in total agreement. A big "aha" I have not considered is your point about the bias being in stream with the speaker. Your rational makes a lot of sense. I also had reached a tentative conclusion that the caps are necessary but more from the standpoint that I wasn't quite tracking NP's progression of the circuit from simple to less simple.
"Will anything you're proposing re: modifications to the circuit alleviate the concerns re: potential DC reaching my speakers?"
Yes, IMHO. I'm still brought back to NP's comment about powering the bias V with a battery. This implies to me that you don't have to tie into the output, i.e. the source pins. It seems that he is powering the bias voltage using the source pins out of necessity, not out of "audio-motive" requirements. Going on logic regarding what bias V is on paper, a low voltage source, I see no reason to tie externally supplied DC to the source pin, as it would be redundant. That leaves only a few options for how to supply the external DC; the option that jumps out is tying in between the gate and ground. That's my current guess on how to do this. I like this setup because you could have a standalone bias supply, adjustable, that you can plug in to various builds like you would a mains supply. So whenever you tinker on new circuits that is one less variable to contend with.
I have made some progress elsewhere on this general concept. Firstly, It's looking like my idea to use Arnon7 to roll my own transformer is not going to cut it. That particular core material is very efficient based on being very thin laminations, .007", but it doesn't quite hold up at high frequency. It's good but not as good as the nickel alloy used in the Jensen. So that little guy is going on the backburner.
Secondly, I have gotten pretty far into the design stage for a heatsink die to use with the dual channel IC. It looks pretty doable for its tiny pad size. They way it works is each pad gets it's own copper billet that expands outward from the pad footprint. The large billet pads provide enough workspace to implement various electrical connections. The copper billet is sunk to aluminum nitride insulation which is then sunk to an aluminum housing that will see direct contact with a liquid cooling loop. Because I don't know if using a dual channel IC is unwise, I made provisions to assimilate standard through-hole ICs of the single channel type. Basically it's a modular breakout for power transistors. After reading through various threads on FETs one more time, the IC I mentioned earlier still looks very appealing. The only drawback I can imagine is if there is a proximity issue that affects sound quality due to isolation problems. Isolation aside, I would think they should be very closely matched; just speculation. At this point the discussion is afield of the build guide so any more on it should probably get moved to it's own thing. I'll build the die first and if that goes well I start a thread about working through the concept and figuring out what works and what doesn't. Here's the die concept in both IC configs.-
Last edited:
I take back my previous assumption on this point. If the speaker shares the same ground path then the capacitor would have to go in. The F6 diagram does not show a ground on the output and I took that at face value without considering the full connectivity of the drivers. Not too familiar with how the various grounding strategies are applied and will have to do more reading. It basically comes down to whether or not that bias supply can be isolated in a way that mitigates the scenario you mention. On one hand it's so easy to just stick in the caps-- but when the entire footprint of the circuit is considered, they add about a third more to the real estate and there is a massive height increase in the z axis. So it's certainly not negligible from that perspective. Not to mention how much more work it is to integrate the connections. Not the end of the world to have them and it's just sort of interesting to think about more than anything.
Not sure what you mean. If you connected out to ground... you'd literally be shorting the output. Maybe I misinterpreted. If so, apologies. The "OUT" is just one end of the connection. The ground is where the other speaker terminal connects. The speaker is the load between the two, and the speaker reacts to the difference in the potential between "OUT" and "GND". To make things more clear in my head, sometimes I even draw lines between the grounds so I can visualize current flow. Makes for a messy schematic, but it helps me to work my way around. If I've given an overly basic review to something completely irrelevant... again... apologies.
That's one scenario, but a few others are mentioned also. I admit to being unsure how it will react in various situations. They may truly not be needed, but as I noodle it through... wondering how... is fun.
Totally understandable.
Same. Thanks for allowing me to ramble and think along with you.
I don't mean short it directly, I mean the entire loop through the speaker. That negative speaker pole ends up somewhere. From what I gather, that connection is to a common ground plane. The M2 schematic is slightly better at showing the completed circuit, but not entirely. If you look at the traces on the M2 board however, it's clear the the negative speaker out terminal ties directly in with the 'common', or 'neutral' (whatever the correct term is) center tap of the power supply. Point is, there would be some shared connection with the external bias circuit. OTOH, maybe with the FET source pins out of the circuit (thx to external bias supply), you don't need to be as militant on the type/size of caps. The M2 separates the ground plane and the Edcor with the tiny tantalum job (also the bipolar slightly downstream). Even that scenario would be desirable. 🤔
Ahhhh! Which schematic for the M2? If you're referring to Mark's schematic in post #1 of the "The diyAudio First Watt M2x" thread, he's illustrating a few direct contact points. Note Nelson's original in post #1 of this thread ... a bit more similar to the F6 schematic in that regard.
https://www.diyaudio.com/community/threads/official-m2-schematic.281520/
I see where you're coming from. P3, P7, P6 "grounds" in Mark's schematic are all still at the same potential. Mark was just nice enough to mark each one separately similar to the PCB. It's just a ground.... same as the other grounds... Same with the F6, the ground points on the PCB are just not marked separately in the schematics we're looking at.
Thanks for clarifying what you meant. I was scratching my head.
We watch/listen to a lot of multichannel bluray performances and after being so happy with how the dual mono F6 sounded on the mains, I built a single monobloc to power the center channel using the Modushop monobloc case with two heat sinks. Electrically this build is exactly half of my dual mono version using all the same parts.
Power supply layout. Left side of photo waiting for mounting of heat sinks, right side will hold a blank panel.
Initial smoke test and bias adjustment.
Construction complete other than top cover.
Some comparisons of a 4U case versus the monobloc case.
Power supply layout. Left side of photo waiting for mounting of heat sinks, right side will hold a blank panel.
Initial smoke test and bias adjustment.
Construction complete other than top cover.
Some comparisons of a 4U case versus the monobloc case.
I have one that I made with cut and paste from posts #1 and #3 in another file format for my own use. I converted it to .pdf. I did not take time to re-format, and whatever images are missing are still missing... but it looks fine. It is too large to post. PM with your e-mail, and I'll gladly send it along.
Of the commonly available thermistors, I prefer the CL-70 for inrush current limiting. Name is easy to remember. It maintains a higher resistance during startup into large capacitors.
The F6 is on the edge of needing an even larger device. There was a separate thread on this very subject (optimal thermistors).
The F6 is on the edge of needing an even larger device. There was a separate thread on this very subject (optimal thermistors).
I am having understanding issues... many here can easily answer these.
- where does the 'SOFT START & SPEAKER TURN-ON DELAY / DC' fit into the flow
- having confusion as to which torrid transformer to order... suggestion as what to order
- any scuttle as to when the F6 kits will be back into inventory
- where does the 'SOFT START & SPEAKER TURN-ON DELAY / DC' fit into the flow
- having confusion as to which torrid transformer to order... suggestion as what to order
- any scuttle as to when the F6 kits will be back into inventory
I can't comment on the soft start or speaker delay as I am unfamiliar with their design.
On the transformer, an Antek 300VA 18V will work just fine: https://www.antekinc.com/as-3218-300va-18v-transformer/
For overkill, can step up to the 400VA or 500VA versions.
On the transformer, an Antek 300VA 18V will work just fine: https://www.antekinc.com/as-3218-300va-18v-transformer/
For overkill, can step up to the 400VA or 500VA versions.