"Proper" TDA3116 PCB

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
cogitech

cogitech

Member
Joined 2011
Paid Member
Thanks guys. I am thinking of building my own linear supply with a relatively sturdy AC transformer, a bridge rectifier and some good caps. I've got all the parts, so figured I might as well do that rather than spend money on a switching supply (unless there was a good reason to).
 
R

rsavas

Member
Joined 2012
Using a linear, calculate for a high line condition, 120V +10 % = 132V although it rarely happens.
I have my portable running off a bench supply 24V with an ammeter, you will be surprised how little power these chips draw.
For unregulated, I suggest a 75-100VA transformer would be more than good enough. If you keep the transformer far away, you get absolutely no hum.
I think you are okay with a 16V trans, a 12.6V is a bit low, but can be used at reduced dynamic range.
16V(rms full load) * 1.414(pk) - 1.4V (full-wave diode loss) = ~21V
A few options
546-186E16
546-186F16
546-187F16

A 10mF(10KuF)/35V ecap
Tons to choose from
Capacitors | Mouser

Can not forget the bridge diode
http://ca.mouser.com/Semiconductors...-ax1mf?P=1z0z1ruZ1yzxpggZ1yzxpgf&Ns=Pricing|0

and a primary fuse too
 
Last edited:
cogitech

cogitech

Member
Joined 2011
Paid Member
rsavas said:
Using a linear, calculate for a high line condition, 120V +10 % = 132V although it rarely happens.
I have my portable running off a bench supply 24V with an ammeter, you will be surprised how little power these chips draw.
For unregulated, I suggest a 75-100VA transformer would be more than good enough. If you keep the transformer far away, you get absolutely no hum.
I think you are okay with a 16V trans, a 12.6V is a bit low, but can be used at reduced dynamic range.
16V(rms full load) * 1.414(pk) - 1.4V (full-wave diode loss) = ~21V
A few options
546-186E16
546-186F16
546-187F16

A 10mF(10KuF)/35V ecap
Tons to choose from
Capacitors | Mouser

Can not forget the bridge diode
Bridge Rectifiers | Mouser

and a primary fuse too
Click to expand...

Hey! Thanks for all that.

The parts I have, and which I tinkered with after work today, are:

- A 40VA transformer with 120v primary and 12v secondary, which measures out at 13.7vac unloaded.
- A random (big) bridge rectifier that I had laying around (among several others)
- A Nippon Chemi-Con 50v 3300uF CEW "big black" (which tests perfectly compared to a few of its brothers I have in a bin here)

After I hooked it all up and let the old cap settle in for a while, I measured a steady 18.5v DC.

I discharged and measured the cap after this and it still comes in around 3200uF, right where it measured before charging and discharging.

This stuff will have to do for now, as I am blowing my budget on Gary's fully-built board. :) I figure I will only be using asking the amp for about 15w on Fridays and Saturdays...

I do have a 19v, 90w HP laptop brick here as well, but there's nothing romantic about using that ugly thing. :D
 
Last edited:
gmarsh

gmarsh

Member
Joined 2004
You'll probably find the 40VA transformer supply will dip a bit if you're playing music hard, and dip a lot (probably with the transformer getting a bit toasty) if you attempt a tone test. But with typical music listening levels with reasonably efficient speakers it'll work just fine.

Giv'er, and see what happens I guess. It's pretty hard to cause damage to anything with the setup you're describing.
 
cogitech

cogitech

Member
Joined 2011
Paid Member
gmarsh said:
You'll probably find the 40VA transformer supply will dip a bit if you're playing music hard, and dip a lot (probably with the transformer getting a bit toasty) if you attempt a tone test. But with typical music listening levels with reasonably efficient speakers it'll work just fine.

Giv'er, and see what happens I guess. It's pretty hard to cause damage to anything with the setup you're describing.
Click to expand...

That's what I figure. Nothing to lose and lots of fun tinkering.

I've been playing with simple linear supply simulations in LTSpice and have seen what a dramatic effect the capacitor(s) specifications have on ripple smoothing. Pretty cool stuff. I have a box in the shed full of HUGE caps (some as large or larger than a can of beer). Some of them are well over 10,000uF if I recall correctly. Hopefully they test OK.

One question though; for modelling the simulations in LTspice, I need a load in order to generate ripple. Right now I am using a simple 30ohm resistor (based on mooly's excellent tutorial). I have noticed, however, that changing the value of this load has a huge effect on the ripple, so I am wondering what is an approriate load value (in ohms) to simulate the load of the TPA3118 amp that you have designed?


 
doctormord

doctormord

Member
Joined 2007
360customs.de
Your nominal speakerimdedance/2 switched by a generator at 400kHz which is modulated by your test frequency. -.- You might want to use Tina-TI with a model of the amp.

To make it easier you might want to use a frequency modulated current-sink/source. (You might run into trouble as the build-in current-source/sink is having unlimited voltage properties)
 
2

29285

Disabled Account
Joined 2010
If you push the full-bridge to its limits, i.e. the sinewave starts to clip, your supply sees a load twice the speaker impedance.
In that case, a full bridge amp loaded with 8 Ohm speaker can be replaced by 16R dummy load.
Considering full overdrive (chaotic party events) the dummy may be reduced to single speaker impedance.
 
cogitech

cogitech

Member
Joined 2011
Paid Member
I should probably have mentioned that I really have very little knowledge of electronics.

Therefore I only understand about 30% of what doctormord just wrote.

I do appreciate the help, though. :)

Even just a "guestimate" of the load would be helpful - assume 4 ohm nominal speaker impedance and 1kHz tone.

P.S. Is there already a Tina-TI model of Gary's amp available somewhere?
 
cogitech

cogitech

Member
Joined 2011
Paid Member
With the dummy load set at 8ohms (twice my speaker impedance) the importance of big smoothing caps becomes very evident. I will definitely be looking through that box of mine to see if those 25,000uF and 30,000uF (Mallory/CDE "CGS") cans test OK. :)
 
Last edited:
doctormord

doctormord

Member
Joined 2007
360customs.de
From real life I can say, that you will be surprised how little power is needed to have an average listening level.

With average 86dB speakers (6-8R) there will be ~25mA quiescent current + 10-15mA per channel. (Iavg)

This was tested at PVCC=5V. :)

So less than 100mW per channel or less than 3mW per Board.

(I actually run a PAM8403 straight from two supercaps 2.7V/25F in series for about 50 minutes)

An externally hosted image should be here but it was not working when we last tested it.
 
Last edited:
cogitech

cogitech

Member
Joined 2011
Paid Member
rsavas said:
From the data sheet,page 6 and figure 24
KSVR Power supply ripple rejection 200 mVpp ripple at 1 kHz, Gain = 20 dB, Inputs AC-coupled to ground
typically = –70 dB
Click to expand...

Yes, I read that and only half understand it. I'm guessing it means that I need to keep my supply ripple at or below .2 volts in order to have -70dB of rejection.

I also assume that this means if I have >.2v of supply ripple, then the ripple rejection will be worse than -70dB.

Obtaining less than .2v of supply ripple with an 8ohm dummy load requires either huge caps or regulation, according to my playing in LTspice.

I am tempted to follow Mooly's LTspice guide and build this linear supply out with a voltage doubler and regulation. I've got a model in spice right now that shows it putting out 25v of perfectly flat voltage. But perhaps I should take baby steps.
 
Last edited:
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom