maybe not effectively sure, but its not gonna be a reliable pass thru of RF energy either, so yes it does throw HF away ( dissipate or radiate take your pick) what it can't use.
it's surely not a good RF ground so its just a part of your nasty antenna!
Last edited:
so yer pitching this card against the Ebay / Chinese race to the bottom.
interesting ..
Last edited:
It's a race to the bottom because it's a card with fixed gain and no UI, intended to be embedded inside a project? Guess that means UcD/nCore amps are **** too...
Excellent gentlepersons! This higher-level, collaborative engineering is exactly what I was looking for; instead of the "I pseudo-randomly changed some caps and hear a 'difference' ..." of the other thread. 🙄
I know this thread has taken on a life of it's own and because of that has exceed all my expectations. 😎
So if you want to tweak a bit more, here's what I was personally looking for.
Full functionality (every setting, feature) quickly changeable by dip switch (or if dip is too big, costly; then just pin headers and jumpers) 🙂
Somehow reading the fault/diagnostic pin. I wanted to 'play' with some features (like the power output limiting function and then be able to see when it's clipping so I could connect it to what I'm hearing ) 😱
So there's my 2 cents. Thanks again to gmarsh for all his solid high-speed engineering work.
Cheers to all,
Jeff
I know this thread has taken on a life of it's own and because of that has exceed all my expectations. 😎
So if you want to tweak a bit more, here's what I was personally looking for.
Full functionality (every setting, feature) quickly changeable by dip switch (or if dip is too big, costly; then just pin headers and jumpers) 🙂
Somehow reading the fault/diagnostic pin. I wanted to 'play' with some features (like the power output limiting function and then be able to see when it's clipping so I could connect it to what I'm hearing ) 😱
So there's my 2 cents. Thanks again to gmarsh for all his solid high-speed engineering work.
Cheers to all,
Jeff
Last edited:
You won't be able to sense clipping easily with anything other than an oscilloscope, FAULTZ won't activate when you're clipping or anything. FAULTZ is accessible on a test point if you want to watch it.
Made a few changes:
- Comparator circuit isolated with a diode, which should fix any "pull the power cord pop" (thanks for the suggestion, doctormord!)
- PLIMIT now has a resistor divider from GVDD, added allowing PLIMIT to be programmed. Bridge the high resistor if you don't want the feature.
- Changed 6 pin input/UI header to 8 pin, added a spot for a LED. LED is lit when the TPA3118's MUTE signal is low, it'll turn off if the amp is shut off and blink off if a fault occurs.
- Renamed OFF input to STBY, which makes more sense 🙂
- Gain/SLV DIP switch now badly labelled.
Pretty much running out of changes to make to the amp now.
Made a few changes:
- Comparator circuit isolated with a diode, which should fix any "pull the power cord pop" (thanks for the suggestion, doctormord!)
- PLIMIT now has a resistor divider from GVDD, added allowing PLIMIT to be programmed. Bridge the high resistor if you don't want the feature.
- Changed 6 pin input/UI header to 8 pin, added a spot for a LED. LED is lit when the TPA3118's MUTE signal is low, it'll turn off if the amp is shut off and blink off if a fault occurs.
- Renamed OFF input to STBY, which makes more sense 🙂
- Gain/SLV DIP switch now badly labelled.
Pretty much running out of changes to make to the amp now.
Last edited:
This process doesn't solder the chip to the PCB thermal pad, so the Theta J is a big variable. IMO eg the chip is cold when pressing down on the PCB
Done in a production oven everything is pre-heated then all is solder bonded in one go.
Last edited:
Yes it does!
You pre-tin the pad on the TPA with solder, and dab some flux on the blob of solder on the PCB. When the blob of solder under the TPA melts when you're applying bottom heat, the unmelted solder on the bottom of the TPA will pretty much immediately melt to it, forming the joint. There's not much thermal mass on the TPA, and having 20mm^2 of molten solder up against the thermal pad will heat it up in a hurry.
You feel the solder melt when the TPA nudges closer to the board. Keep the heat on for a few more seconds, to make sure everything's nicely flowed together and the flux is burnt off, and you're good. I've done this many times for board level switchers, high speed data converters, etc... never had any problems.
You pre-tin the pad on the TPA with solder, and dab some flux on the blob of solder on the PCB. When the blob of solder under the TPA melts when you're applying bottom heat, the unmelted solder on the bottom of the TPA will pretty much immediately melt to it, forming the joint. There's not much thermal mass on the TPA, and having 20mm^2 of molten solder up against the thermal pad will heat it up in a hurry.
You feel the solder melt when the TPA nudges closer to the board. Keep the heat on for a few more seconds, to make sure everything's nicely flowed together and the flux is burnt off, and you're good. I've done this many times for board level switchers, high speed data converters, etc... never had any problems.
thanks ah I see better so the chip is actually in contact when the heat is applied 1st.
Im used to using heat plates on bigger things alumina etc, not SMD FR4
Im used to using heat plates on bigger things alumina etc, not SMD FR4
Last edited:
Nice work Gary! Any idea on how much it will cost? Might buy one just to know what proper enginereering sounds like ;p What's the recommended PSU?
I don't have a recommended PSU, I'm sure there's other people on here and on the TPA3116 thread that have made various recommendations. Normally I reach for a Mean Well supply, just be careful buying them off eBay as they're often counterfeited.
I'm gonna largely copy DUG's PBTL group buy:
$8 for a bare PCB
add $7 plus chip cost for install/test of a TPA3118
add $25 plus parts cost for full SMT assembly or full PCB assembly, including test. My board's got a fair number more parts than DUG's to solder, so $25 instead of $18.
Parts kits will be available for the through-hole parts, as well as standoffs/hardware and thermal interface pads if you want to mount the PCB to a metal enclosure with good cooling. Kitting SMT is a pain in the *** so I won't offer that option, I'll let Digikey/Mouser/whoever do that for me 🙂
Customization (using different parts than my own BOM, leaving parts off, etc) is free, you'll pay the parts cost of the parts that go on the board. I'm making orders from Digikey and Mouser to build this set of cards, parts must come from those sites, or be shipped to me.
Shipping isn't included. Bare PCBs and those with SMT will be shipped in a static bag in a padded envelope, full boards and those with parts kits will be shipped in a box. I'm shipping from Halifax, NS via Canada Post, prices should be comparable to DUG's shipping costs but may be slightly more as he's over in Ontario.
I'll be open with my parts costs. I'll be "taxing" them to account for shipping/duty costs to get them here, 15% HST, and test hardware that I have to build (2 bed of nails testers - one for TPA only, one for full SMT, probably $50 for the two)
All prices are in canadian dollars, payment will be taken via interac e-transfer or paypal.
Again, if all functions could be DIP or pinheader selectable it would be much easier, faster, better for A / B testing, better for newbs, less hassle (for end user) ... 🙂
Thanks so much gmarsh,
Cheers,
Jeff
May I suggest to load the bom into Mouser project manager, so that they do the kitting. I have also added notes so that if a comp is back-ordered you can provide alternates if they exist. You can then share the bom. I will do it for you, if you send me the bom, there are not that many parts.
Another suggestion is to use a bigger via fhs in the middle of the power pad. See attached, I used a 100pad/75 mil fhs, I'll post a pic of what I am talking about.
Another suggestion is to use a bigger via fhs in the middle of the power pad. See attached, I used a 100pad/75 mil fhs, I'll post a pic of what I am talking about.
Last edited:
Everything's DIP switch selectable with the exception of the PLIMIT threshold, which is set by changing resistor values. If you want something quick to adjust, you can bodge on a potentiometer.
Definitely. I've considered building a toaster oven reflow system, how's it working out?
That's the plan!
I'd rather keep it the way it is with lots of small vias, just found out my board house allows 0.3mm vias so I'll probably cram even more in. Copper's thermal conductivity is 10x that of lead.
I only have one point to say about the via barrier from top gnd to bottom. It will act as a heat transfer barrier as well, preventing an effective heat transfer on the right side. Mr. Adam from Adam research did a lot FEM simulations to show this problem. If you need further informations, I'll have a look for the publications.
So will take at least 8 pcbs, probably 10. Probably all with the chip soldered. Affordable the full SMT build, do you expect them to be done by hand or via a pick and place with reflow?
Referring to the lines of ground vias to the right of the decoupling caps? Should They should be OK, they don't really block the direction of the heat flow at any significant angle.
There's really no need for me to put so many of them down, but hey, drills are free 🙂
They so because they are so close to each other.
Page 7:
http://www.ti.com/lit/an/snva419c/snva419c.pdf
"Thermal Break" or "Thermal Breakdown"
Page 7:
http://www.ti.com/lit/an/snva419c/snva419c.pdf
"Thermal Break" or "Thermal Breakdown"
Last edited:
It works awesome and exceeded all my expectations.
This was my first project on it. And more here. Details on oven construction can be found elsewhere in that thread.
The arduino shield is from Rocket Scream and seems to be bomb-proof.
Combined with good solder paste and cheap stencils it's a winner. You spend some time placing components, but the actual soldering is push-button and perfect.
BK
- Status
- Not open for further replies.