Thanks a bunch!
Seems half the people that want this amp are canadian. What's with that, eh?
Please keep in mind that shutting down should be accomplished by inserting a switch in series with the 78L05 regulator.
This greatly reduces standby current.
I pulled Mute up to my Vcc(3.3V) with 10K, for the mcu port.
I suggest:
PB3 or PB4, Connect to TPA mute pin with a 10K pullup.
Ports power up in input state or are hi-Z in reset. You can then connect sdz, other inputs, to the shared programming port pin (MOSI,MISO,SCK). You can put the optional series R's but I did not have them.
Option: Diode or zener (1W) value set by 78L05 V I/O max - 5V, across I/O of 78L05. Could also have a 5.6V zener at 78L05 o/p to ground, a crude OVP ckt
Ferrite bead in series with 78L05 i/p &/or R/C/R to filter out possible mcu switching noise, transfer heat dissipate form 78L05 to series R's.
28V(max)-5V * 6mA(max @8Mhz/5V) = 138mW, okay for the SOT-89 I think, have to check
Last edited:
Latest.
I'm gonna disable RST to keep MUTE in a defined state with /STBY pulled low. I honestly don't have the room for the pull-up
SDZ is tied to PB3 (one of the non-ISP pins) with a 30K pulldown, which will keep the TPA off during programming. MUTE, FAULTZ and the PVCC sense divider are the three ISP signals and twiddling them should be OK with SDZ low.
I'll probably run the AVR off its 128K oscillator to save power, I don't need much speed or accuracy. Programming happens pretty fast and I think the 7805 (or MCP1804 or whatever I shove in that spot) should have enough thermal mass to handle it, these regulators all have thermal protection anyway.
And I wouldn't worry about the AVR clock coupling back through the regulator - considering the effective series R of the regulator and the amount of decoupling on PVCC probably makes a lowpass in the mHz or uHz, I doubt you'll get any measurable let alone audible ripple.
It's kinda like worrying about windshield wiper noise on a car with uncapped headers.
Take your time my friend.
The Wiener is a slow-cooked delicacy, turning over and over and basting in its own juices.
It's allllllmost done. No need to rush it at the very end.
(Hey, the cooking analogy is pretty apt! 'over and over' like this thread's mulling (good) of the bypassing, functions; and 'basting in juices' working with the startup, powerup, ucontroller, ... )
Well here you go. It's officially The Wiener, and I hope the long process and expensive list of ingredients will yield a delicious result in the end, without any smoky flavor.Take your time my friend.
The Wiener is a slow-cooked delicacy, turning over and over and basting in its own juices.
It's allllllmost done. No need to rush it at the very end.
(Hey, the cooking analogy is pretty apt! 'over and over' like this thread's mulling (good) of the bypassing, functions; and 'basting in juices' working with the startup, powerup, ucontroller, ... )
So all my ideas are not the best, at least I am contributing!!
"gmarsh" shows class, a fine Canadian, which you two seem to lack and he also gives a valid reason why it is not necessary or worthy of inclusion.
You guys are akin to a bunch of bullies that want to kick someone, when they are down. GROW UP CHILDREN!!
I actually take offense to unnecessary statements such as these as they add constructively NOTHING to the conversation.
"gmarsh" shows class, a fine Canadian, which you two seem to lack and he also gives a valid reason why it is not necessary or worthy of inclusion.
You guys are akin to a bunch of bullies that want to kick someone, when they are down. GROW UP CHILDREN!!
Last edited:
I'm very interested in this and how it was "cooked". Now may I know what it would probably cost, so I can start counting the pennies in the piggy bank.
It just occurred to me, DFM review
I have no idea about your mfg plans, but if you ever decide that you want these pcb's sent out to a sub-contract mfg, for assembly, you might want to put on 3 imaging fiducials, for the pick and place machine registration/alignment.
These are usually put in 3 corners, of the pcb, on each side SM comps are mounted on. Clear of traces underneath.
The global fiducials can be put underneath TH comps, as they usually go on after reflow.
Another DFM rule was to use one or two fiducials, for each fine pitch sm comp, if pitch was < 50 mil. One in the middle or two on opposite corners of a qfp.
Not sure if they can image, using the via smack in the middle of the via array underneath the TPA. Have to ask the CMfg on that one.
You have 4 tooling holes (might not be the preferred hole size) for a stencil alignment, so you are good there.
You have a few TH comps, so the mfg would decide if they do a wave or hand back load , it really depends on the qty involved. There was also another option where they did a selective wave process rather than hand soldering which adds cost, even if it is done in Asia.
I have no idea about your mfg plans, but if you ever decide that you want these pcb's sent out to a sub-contract mfg, for assembly, you might want to put on 3 imaging fiducials, for the pick and place machine registration/alignment.
These are usually put in 3 corners, of the pcb, on each side SM comps are mounted on. Clear of traces underneath.
The global fiducials can be put underneath TH comps, as they usually go on after reflow.
Another DFM rule was to use one or two fiducials, for each fine pitch sm comp, if pitch was < 50 mil. One in the middle or two on opposite corners of a qfp.
Not sure if they can image, using the via smack in the middle of the via array underneath the TPA. Have to ask the CMfg on that one.
You have 4 tooling holes (might not be the preferred hole size) for a stencil alignment, so you are good there.
You have a few TH comps, so the mfg would decide if they do a wave or hand back load , it really depends on the qty involved. There was also another option where they did a selective wave process rather than hand soldering which adds cost, even if it is done in Asia.
Last edited:
Today I did some investigations on turn-on/turn-off plopping of TPA32xx. So I built a simple Attiny25 controller, flashed some power-up/power-down sequence into it and connected the ports to mute and SDZ.
On the other hand, I wanted to enforce plopping, so I connected one analog input via 100uF+10k to GND, the other left open. This gave considerable plopping, specially at power-down.
And the result is, the Attiny25controller did NOT suppress that anyway!
So, what went wrong? -
I will give some ideas
Whenever you hear any noise from the amp, there is a voltage differential between the two legs of the h-bridge. So, if voltages of both legs build up exactly the same during power up and power down the differential cancels and there will be no plopping at all. But if there is some plopping this is an indication of asymmetric behaviour of the h-bridge, i.e. some kind of mismatch between both half-bridges.
This became evident with the prototype of doctormord which I could check here. Plopping was at its max with the 10kOhm input pot in center position, but vanished in the end positions. The explanation is simply an increased series resistance on one input, while the other input is grounded via the same coupling cap but no series resistance. This is an unbalance that gave rise to plop noise.
Ok, never mind, lets do some sequencing around mute and SDZ and everything will be fine...
Nope! Sequencing requires a way of noiseless switching - and that is the problem...
With unbalanced inputs, toggling mute or SDZ input level even with stable supply voltage results in plopping, so there is no way to switch off noiseless in case of unbalanced inputs.
Conclusion
I do see no point in any sequencing of SDZ or MUTE inputs – as long as inputs are unbalanced. Any focus should be on perfect input impedance balance.
On the other hand, I wanted to enforce plopping, so I connected one analog input via 100uF+10k to GND, the other left open. This gave considerable plopping, specially at power-down.
And the result is, the Attiny25controller did NOT suppress that anyway!
So, what went wrong? -
I will give some ideas
Whenever you hear any noise from the amp, there is a voltage differential between the two legs of the h-bridge. So, if voltages of both legs build up exactly the same during power up and power down the differential cancels and there will be no plopping at all. But if there is some plopping this is an indication of asymmetric behaviour of the h-bridge, i.e. some kind of mismatch between both half-bridges.
This became evident with the prototype of doctormord which I could check here. Plopping was at its max with the 10kOhm input pot in center position, but vanished in the end positions. The explanation is simply an increased series resistance on one input, while the other input is grounded via the same coupling cap but no series resistance. This is an unbalance that gave rise to plop noise.
Ok, never mind, lets do some sequencing around mute and SDZ and everything will be fine...
Nope! Sequencing requires a way of noiseless switching - and that is the problem...
With unbalanced inputs, toggling mute or SDZ input level even with stable supply voltage results in plopping, so there is no way to switch off noiseless in case of unbalanced inputs.
Conclusion
I do see no point in any sequencing of SDZ or MUTE inputs – as long as inputs are unbalanced. Any focus should be on perfect input impedance balance.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.