Very low power (< 1W) amp project and questions

[Baptiste]

Ok, first decision made, I'll go for the 6au6. A single one for now ;-)
If you have any recommendations about brands (preferred or to avoid), whether it's important or not... I'm listening.

Thanks for your feedbacks about the OT. It seems there is not really a wrong choice.
lf the Hammond has no other drawback aside from the price, I'll probably go for it for the added flexibility. In Europe it's about 34€, vs 17€ for the reverb one.

[Baptiste]

Hi everyone,

I’ve been working a bit on my design.
Here is a first draft of the power section schematic (nothing fancy here), and the power supply (more interesting).
I didn’t know much about power supply design before this exercise, I learned quite a few interesting things in the process.
For those who seek some relevant tutorials about power supply design, here are the few I used :
-   http://diyaudioprojects.com/Technical/Tube-Power-Supplies/
-   http://www.valvewizard.co.uk/smoothing.html
-   https://dalmura.com.au/static/Hum%20article.pdf

And I used the PSU designer 2 software from Duncan Amps:
-   http://www.duncanamps.com/psud2/

Design considerations:
-   First design based on readily available transformers => 190-200V AC secondary voltage
-   150V DC for power stage
-   200-250V DC for preamp stage (12ax7 in mind)
-   VVR for power stage
-   Standard values for capacitors

I tried to factorize filtering between power and preamp stages, and to limit the size of the reservoir cap. Pretty easy considering the low current and voltage required for the power section. With the current values, it’s already quite overkill, but 10uF capacitors are cheap.
Ripple voltage is around 3mV for V1, 40mV for V2, 130mV for 6au6 anode, 100mV for G2.

Interesting fact is that it seems I can make a poor man’s VVR with a simple potentiometer, since I have very low power dissipation. It seems too simple to be correct. Did I miss something ?

Other noticeable fact is a heavy coupling between R3 (anode voltage drop resistor) and R8 (cathode bias resistor): due to the important value for R3, a tiny adjustment on current bias has a big effect on the 6au6 voltage.

I will review components values once I’ll be settled on a power transformer.

Any thoughts ?

[sluckey]

Looks like you will expect 14mA to flow through that 50K VR1 pot. The pot will have to dissipate 9.8 watts when set to max resistance. 

[Baptiste]

No, the current drops as the voltage drop increases on the pot. At 50k, there would be almost no current. 50k is probably too much btw, maybe 25 would be better, but it depends on the bias...

At 100v reduction through the pot, I would get only 3-4mA.
At 50v, maybe around 8mA.
Power dissipation is under 1W in any case :-)
I found 3W 500V Bourns pots that would seem to be well suited for this.

[PRR]

> the current drops

Not enough for a low-price pot. The load is 150V at 14mA or like a 10.7k resistor. This will vary some at lower voltage but not a huge amount. Do the math. I figure the 50k extreme may need a $7 pot. But the worst-case is surely at part-turn where only PART of the track is taking the heat. Dart-boarding 13k (makes math easier) gives 0.6 Watts in 1/4 of the track. Conservatively this wants a 2.4 Watt pot. Yes, the mil-type "2W" part will handle it in non-critical duty (where you do not have 20,000 adoring fans to disappoint).

I'm no big fan of crystals in vacuum-state audio, but this does call for a MOSFET. (Heck, I've done it with a TV H-sweep NPN.)

[PRR]

Let the idiot work the numbers. For these values, pot dissipation peaks at 18k which we could have figured from basic principles. However that heat is released in 18/50 or 0.36 of the pot track. So conservatively a 1/4W pot can take 0.25*0.36= 0.09 Watts. But it is dissipating more like 0.9 Watts in that third-track.

The advantage of the idiot is that it will do this a thousand times and make a plot. When we figure Watts divided by track-in-use, we find the per-area dissipation SOARS as we get toward zero Ohms. I bet at some point in here even a mil-spec pot will go "poof" and the first 1/10th turn is "dead".

[DummyLoad]

https://www.mouser.com/ProductDetail/CTS-Electronic-Components/026TB32R253B1B1?qs=sGAEpiMZZMtC25l1F4XBU7zZBtuyelZnQyjH9uc03iI%3D


RV4 (2W) would probably work, but heat up nice: you really need a wirewound pot. the above is a 5W WW, but only available up to 25K - use 27K 3W R in series. the next step up is a 10W rheostat that are available in 50K value, but fairly expensive and difficult to find stock.


--pete

[Baptiste]

> 0.6 Watts in 1/4 of the track. Conservatively this wants a 2.4 Watt pot.

Indeed! I completely overlooked the fact that .
Many thanks PRR for the graph.

I searched for info about power dissipation of potentiometers, assuming it must be non linear, given the dissipation surface at disposal when the pot is full open.
https://www.bourns.com/pdfs/OnlinePotentiometerHandbook.pdf

On this resource, it states (p44 of the pdf) that for metal-encased pots only, the admissible power dissipation is indeed non linear, and allow to dissipate ~50% of the power rating on 20% of the resistance (2.5 ratio). Asymptote at the beginning of the curve gives a power rating ratio of about 3.
This is not true for plastic encased pots.

If I consider PRR's calculations, with 14mA at 1ohm, it dissipates about 0.2mW versus 0.2 to 0.6 mW allowed for a 25k 5W pot.

So, a pot such as the one DummyLoad proposed should be compatible, but without much power margin. And the 25k max value is a bit limiting (ideally I'd prefer to have ≥35-40k)

Other options :
- Mosfet based VVR => I'll go this way if needed, but I'd rather try the simpler options first.
- Switchable resistance instead of VR1 => I don't need fine tuning of the voltage reduction, so a single switch or a 3/4 position rotary selector would be enough.

I think I will go for a switchable resistance, cheaper, more reliable and not limited to 25k.
I suppose the voltage/current discontinuities when switching the resistor will be handled by the RC filter capacitor.
Do you think of other caveats with this solution ?

[shooter]

Let the idiot work the numbers

it was a crayon night, but I felt embolden so I used a pencil 

[PRR]

> switchable resistance

Makes sense. How often will 0.33W be spot-on but 0.32W or 0.34W gets you thrown out of the club (bedroom?)? Taking 3dB-4dB steps over a 15dB-20dB range (all a single tube can cover) suggests a 6-way switch which is about $2. (You are over the voltage rating, so buy a spare, and don't switch directly into a capacitor without some series resistance.)

I think you have 'mW' a few places in #57 which should be 'W'.

[DummyLoad]

https://www.ebay.com/itm/De-Jur-Amsco-Radio-Rheostat-50K-Spec-No-KS-9004-NOS/293468475051?hash=item4454158eab:g:z5kAAOSwYpxeQIOT


--pete

[Baptiste]

You are over the voltage rating, so buy a spare

There shouldn't be much over 100V across the selector. Or do you mean the absolute voltage if the (plastic?) selector is grounded ? I could start with putting it after the dropping resistor to lower the absolute voltage at 150v.

don't switch directly into a capacitor without some series resistance.

Could you explain a bit further, I'm not sure to understand ? Is it to limit peak current when switching ? I have the dropping resistor in series with the switch already. Is it enough ?

I think you have 'mW' a few places in #57 which should be 'W'.

No that's what I meant. 14mA across 1 ohm gives 0.2mW, versus 1/25000*5W = 0.2mW theorical power limit. Values are ridiculously low at such resistance values.
(BTW, I can't picture burning a 5w pot with such a low power, even if it's only across 1/25000 of the resistor...)

https://www.ebay.com/itm/De-Jur-Amsco-Radio-Rheostat-50K-Spec-No-KS-9004-NOS/293468475051?hash=item4454158eab:g:z5kAAOSwYpxeQIOT
--pete

Impressive beast ! I won't go that route, but good to know it exists, and at such prices ;-)

[Baptiste]

Hi everyone,

Time flies, isn't it ?
I've been quite occupied with other topics these last months, but I found some time to buy the components for my project, and I did the first build. So here is a recap !

So, I made a quick prototype based on the power supply and 6AU6 power section schematics I posted earlier, with a champ 5F1 preamp, to have a simple first working version. I only use 3 knobs : volume, master volume, power scaling. The 3 others are for the future tone stack.



I'm quite happy, it worked straight away, despite the quick and dirty method I used for prototyping. It's a bit buzzy, not a big surprise given how I did not care about lead dress (except for heaters). It's still perfectly usable at low volumes though, and it didn't bother me enough to redo it yet. 

PSU and power scaling design
So far I'm very happy about that, the PSU works fine, and the SE 6AU6 seems to be the right power for my needs. I used a 6 position selector for power scaling, which changes the dropping resistor for the power section only (anode and G2). 
The range of values I selected give me from 50V to 150V for the 6AU6 anode. At max voltage the 6AU6 provides about 0.4W, and I have enough headroom for my needs. At lowest voltage, It gives around 50mW, perfect for late night practice.
I did not try to get lower in voltage, maybe I could go down to 40V, but I suspect I might be near the reasonable limit, because at 50V I'm already really low on grid bias (-0.55V, versus -1.8V @150V).

Given the very low output power, the dropping resistor handles the power dissipation by itself, and I don't see the need for any voltage regulator. 
There is a caveat though with the "dropping resistor power scaling" that I didn't expect: at startup, the dropping resistor is uneffective, because the 6AU6 draws no current while it's still cold. For 10 seconds, the tube sees the full PSU voltage on anode an G2, about 315V (which is much higher than what I expected).
From what I read, over-voltage at startup seems like no big deal for most tubes, but I don't really know for the 6AU6, it's twice the maximum voltage for G2. Anyway, the mighty 6AU6 survived a couple hard starts, and since I have my 6 position selector at hand, I decided to use the first position as a standby switch. This standby only affects the power tube and it works fine. 

About the build 
At first, I wanted to fully breadboard the amp, with a convenient way to test and change conponents and connections. I searched for ideas, and I found these Chinese spring connectors:
https://a.aliexpress.com/_dTbctRZ

I first planned to use them flat on a large wooden panel, along with the transformers, tube sockets... I changed my mind, thinking it would be safer and faster to use a real chassis. For the sake of simplicity, I used a small pre-punched chassis from Tube-Town.net. I think it was the right move, it took me maybe 4 hours to fasten everything mechanical (spent some time on a plexiglass support for the knobs), and 2 evenings to prototype and power on.


spring connectors


soldered ground bus

In the end, I'm a bit mitigated with this connectors approach, mostly because I didn't tweak much components values after the build. However, I think it could be useful for people prototyping a lot. Here are my thoughts about it:

Pros:

• allows prototyping without soldering (useful if you want to advance your breadboarding with young kids around...)
• robust connections so far. Didn't have any connection problems.
• same spacing as the turret board I plan to use after.
• dirt cheap. A few bucks for 100 connectors.
• easier to switch components or correct mistakes, or redo everything.

Cons:

• not much faster than soldering. Might even be slower
• wires must be the same size if you want to put several together
• grounding bus needs soldering anyway (it's a PITA to daisy chain small wires with the connectors without soldering)
• larger than an optimized turret board design => hard to fit in a small chassis.

Next
At first I wanted to experiment with various 3 to 4 gain stage preamp topologies and tweak my own resulting schematic. However I have to admit I don't have the time for this, so I'll try to focus on a more straightforward solution. 
I'd like to go for a well known and versatile 4 gain stage topology, make a clean build for it, and limit myself to minor tweaks after. I spent some time looking for schematics a few months ago, found quite a few here and elsewhere.  I'll make a recap, I'd like to have your advice on these options. If you have any preamp schematic in mind, dont hesitate to share !

Thanks for reading me,
Baptiste