Author Topic: Texas Instruments 3116D (ClassD) amp chip  (Read 10485 times)

0 Members and 1 Guest are viewing this topic.

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #25 on: August 17, 2020, 02:16:27 am »

Yes, they are talking about "max" power output, so sure, you can get 200 watts out of a 24V RMS power supply rated at 5A.

Back to my original problem, I'm not seeing how you get anymore than 100 watts continuous power at 4 Ohms using that 24V/5A power supply.

I think some of my problem is that I don't see where contious power or peak power is stated int eh Texeas Instruments spec file:
Table at section 6.6 of the datasheet indicates 50W continuous at 4 ohms with 21Vcc.  According to the description, this is per channel - in BTL it will do 100W under the same conditions. A side note: keep in mind this is under 10% of signal distortion - far from a hi-fi under these conditions (it will have less distortion if driven at lower power).

So you can get 100 watts @ 20V and 5A and 4 Ohms from a 24V/5A continuous power supply.
The output power equals the supply power minus losses - in other words, the efficiency of the circuit. 24V/5A will give you 120W of supply, which may be just enough for continuous operation of 100W total or 50W per channel - the efficiency is shown at figure 23 of the datasheet.

Higher peak power may or may not be attained. Section 7 describes the drivers will pull up to 7.5A at peak conditions (which gives a 200W peak under 4 ohms).

Hey, thanks for chiming in.

Just so I can understand the technical language, where in that table at 6.6 does it say "per channel" at the 10%THD row?

Power to efficiency are understood. It's a good point, but for now I'm just keeping it at theoretical output for simplicity.

I just want t get the continuous power aspect down given the power supply of 24V at 5A @ 4 Ohms.

If I understand you correctly, this all mean that with two 3316D chips using a 24V / 5A power supply, bridged at 2 Ohms each, for a total, again, of 4 Ohms, they can get theoretical power of 100 watts continuous at 10% THD?

If so, we can move on to the problem I have about 8 Ohms loads and using two of these chips. That would be about 72 watts continuous from the power supply. Well, I'll jsut sate it again:

At 8  Ohms and 24V / 5A = 72 watts.

That's as much power as it can make.

This graph shows a single 3116D chip can do 80 watts continuous. However, the PSU is only going to give it 72 watts.

At .1% THD it will do 66 watts total.

Again, the graph says "maximum power" and says nothing about peak or continuous. I don't know what "Maximum Output Power" means: Peak/Continuous? In any event, if you are running ONE 3116D with a 24v/5A PSU, you get 72 watts total. If you use 2 3116D amps, your continuous output is still going to be no more than 72 watts @ 8 Ohms. I'm only talking about continuous power from the dual 3116D amp chips at 8 Ohms. So what is the benefit of having two chips if you power does not increase over 72 watts, and one chip can use all of it,a nd what's more, at less than .1% THD? I assume you could have even lower THD if both amp chips were putting out a total of 72 watts, but each one was only being driven at 50% power?
« Last Edit: August 17, 2020, 02:18:45 am by DW1961 »
 

Online rsjsouza

  • Super Contributor
  • ***
  • Posts: 6107
  • Country: us
  • Eternally curious
    • Vbe - vídeo blog eletrônico
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #26 on: August 17, 2020, 03:15:30 am »
Just so I can understand the technical language, where in that table at 6.6 does it say "per channel" at the 10%THD row?
The table itself does not say it. The description at sections 1 and 3 says the device puts out 2 x 50W continuous and section 7.4.1 says in PBTL mode (mono) it puts out 100W. With all that, I infer the information there is related to a single channel. 


Power to efficiency are understood. It's a good point, but for now I'm just keeping it at theoretical output for simplicity.

I just want t get the continuous power aspect down given the power supply of 24V at 5A @ 4 Ohms.

If I understand you correctly, this all mean that with two 3316D chips using a 24V / 5A power supply, bridged at 2 Ohms each, for a total, again, of 4 Ohms, they can get theoretical power of 100 watts continuous at 10% THD?
If I understood correctly your amplifier has two devices in PBTL (mono) mode and each is driving its own 2\$\Omega\$ loudspeaker. If so, figure 22 tells me that, at 24V, it can go as high as 150W each device, rendering your 24V/5A rather inadequate to supply this "beast". To get absolute maximum continuous power, you would need to have a minimum of 300W + efficiency losses.

If so, we can move on to the problem I have about 8 Ohms loads and using two of these chips. That would be about 72 watts continuous from the power supply. Well, I'll jsut sate it again:
So, why are you talking about 2\$\Omega\$? Sorry, I really don't get exactly the circling around.

At 8  Ohms and 24V / 5A = 72 watts.

Also, figure 13 says the maximum power at 24V is 42W for 8\$\Omega\$. Given the graph is

That's as much power as it can make.
24V * 5A = 120W. Where did the 72W come from?

With all that, two devices in PBTL mode driving 8\$\Omega\$ should be properly driven by your 24V/5A power supply, given the efficiency would still be well within the tolerance of your power supply.

Again, the graph says "maximum power" and says nothing about peak or continuous.
The characterization having THD+N measurements suggest to me this is continuous - distortion measurements in peak signals is quite difficult to obtain and not very practical.
Vbe - vídeo blog eletrônico http://videos.vbeletronico.com

Oh, the "whys" of the datasheets... The information is there not to be an axiomatic truth, but instead each speck of data must be slowly inhaled while carefully performing a deep search inside oneself to find the true metaphysical sense...
 
The following users thanked this post: DW1961

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #27 on: August 17, 2020, 06:17:46 am »


--------------------
"If I understood correctly your amplifier has two devices in PBTL (mono) mode and each is driving its own 2\$\Omega\$ loudspeaker. If so, figure 22 tells me that, at 24V, it can go as high as 150W each device, rendering your 24V/5A rather inadequate to supply this "beast". To get absolute maximum continuous power, you would need to have a minimum of 300W + efficiency losses. "

I dunno. That's why I'm trying to figure out if how the magic in these chips works.

The single 3116D2 chip runs in BTL mode, which tells me it is really 4 amps on one chip. In BTL mode, you need two chips running mono. However, the 3116D chip runs in stereo. So if it is running in BTL mode all of the time, and it is stereo, then it must have 4 amps on the chip. That's how they can run x2 3116D chips in mono and get stereo in the amp I have, I assume. Otherwise, if it were just two amp chips, they would be only getting a mono amp out of it. So the way I understand it, to get stereo from a BTL system, you need 4 amps.

So the amp I have must really have 8 amps on it (x2 3116D2 chips @ 4 amplifiers each), but they are running the 3116D in mono mode, so they need another one to do stereo.

How they are connecting them together, I have no idea! PBTL? I just dunno.

"If I understood correctly your amplifier has two devices in PBTL (mono) mode and each is driving its own 2\$\Omega\$ loudspeaker. If so, figure 22 tells me that, at 24V, it can go as high as 150W each device, rendering your 24V/5A rather inadequate to supply this "beast". To get absolute maximum continuous power, you would need to have a minimum of 300W + efficiency losses."

As far as power output goes, the chips have a max voltage each of 26V, "Wide Voltage Range: 4.5 V to 26 V" (page 1).

"So, why are you talking about 2\$\Omega\$? Sorry, I really don't get exactly the circling around. "

To get 100 watts RMS from one 3116D, it has to run at 2 Ohms in mono.

"24V * 5A = 120W. Where did the 72W come from? "
Using this calculator: https://www.rapidtables.com/calc/electric/watt-volt-amp-calculator.html
Add resistance at 8 ohms: 8 ohms/3 amps/24v/72 watts. You could increase current, but then you would need to increase voltage too. I guess Ohms law?

They say in the spec sheet that you can get 50w x2 at 21V and 4 ohms.
Using this calculator: https://www.rapidtables.com/calc/electric/watt-volt-amp-calculator.html
It spits out:
4 ohms
5 amps
20 volts

So, ok right on. But, more confusion now. On the right side of the same page, it says:
'The TPA 31xxD2 series are stereo efficient digital amplifier power stage for driving speakers up to 100W / 2Ωin mono."

Page 1

-- 2 × 50 W Into a 4-ΩBTL Load at 21 V (TPA3116D2)
-- The TPA 31xxD2 series are stereo efficient digital amplifier power stage for driving speakers up to 100W / 2Ωin mono.

I have two confusions here.
Using this calculator: https://www.rapidtables.com/calc/electric/watt-volt-amp-calculator.html

(1) 100 watts at 2 ohms:
2 ohms
7 A
14 V
I guess it's ohms law that prevents you from increasing voltage and decreasing amperage to get the same power. If true, then you need a 7 amp PSU to get the 100 watts at 2 ohms in mono.

And I'm assuming again that the 100 watts figure at 2 ohms mono is RMS.

My brain is fried today. I'm still not understanding how they benefit from using two of these chips with a 24v/5A PSU taking into consideration RMS only.

-----------

I found this on the TI website forum and discusses exactly what we were discussing with feedback from a TI rep about RMS and channels etc. Then add that bridging two amps doubles the ohms, so now I'm even more confused than I was earlier today.

"In section 6.6 output power is specified for a 1kHz sine. So, the output power specification is RMS. The output power of 50W is valid for a 4R load in stereo configuration. The TPA3116D2 can drive 2 x 50W into 4R. The TPA3116D2 can also drive 100W into 2R mono, like shown in section 7.4.1 of datasheet (figure 36)." https://e2e.ti.com/support/audio/f/6/t/688690
« Last Edit: August 17, 2020, 06:25:20 am by DW1961 »
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #28 on: August 17, 2020, 01:25:38 pm »
So far, the discussion has related to definitions, arithmetic, and careful reading of the data sheets.
Do you have the basic equipment (audio generator, oscilloscope, AC voltmeter, and 4 ohm power resistor) to measure any of this?
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #29 on: August 17, 2020, 05:05:55 pm »
So far, the discussion has related to definitions, arithmetic, and careful reading of the data sheets.
Do you have the basic equipment (audio generator, oscilloscope, AC voltmeter, and 4 ohm power resistor) to measure any of this?

You forgot knowledge to use the equipment, too.

You've done more than necessary, Tim, and, I appreciate it. I understand if you no longer want to participate in this discussion. It's probably all really mundane for you :(

It just seems like there should be a way to deduce general power output given knows in the spec sheet/diagrams, driver, and resistance. We did that for one chip and came to some conclusions. It was easy after understanding the terminology they were using for "max" power, per channel, etc., which is confusing the way they worded everything--and confirmed by the thread at TI forums.

Now I'm trying to figure out the same thing with another amp that uses x2 3116Ds.

Still not understanding how they get any more power at 8 ohms using two chips, instead of one (which puts out 100 watts total RMS at 20V and 5 amps @ 4 ohms). That's pretty much the max behind the 24v / 5A power supply they provide.

And, again, I'm talking about RMS not peak.

I did get a reply back from the seller that said continuous power is 50 watts per channel at 8 ohms. I doubt it, since:

8 ohms
3.5 amps
28 volts
= 100 watts

I mean they are almost there, but it's a 24 v PSU, not a 28V. Max power for a 24V PSU is 72 watts at 8 ohms (Given the calculator).

Just another quick question, and a quick, simplistic answer, that is bothering me and I've searched for it but no love: Why can't we decrease voltage and increase current (w/o changing resistance) when dealing with resistance in order to get the same power? I know it has to do with Ohms law, and the calculators for ohms and power do not allow trading volts for amps and vice versa, w/o changing ohms too.

By a quick answer, I mean am I right about that? If so, if you want, you could explain it very simplistically. I don't need to know all the physics behind it. Just knowing that you cannot trade volts and amps w/o changing ohms around when dealing with resistance would suffice.
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #30 on: August 17, 2020, 05:56:27 pm »
To your quick question:  to change the ratio of voltage to current (i.e., the load resistance) without changing the DC supply and active devices, a transformer is needed to keep the same power.  A "step-up" transformer will give more volts and less amps into a higher resistance, with the same power (ignoring  transformer losses).  If "N" is the turns ratio, then V2 / V1  =  N  =  I1 / I2  (note the opposite order for V and I),  R2 / R1  = N2 , and P2 = P1.
Solid-state designs avoid transformers, and are optimized for one load impedance, although they will function over a reasonable range of load.  I remember a solid-state linear amplifier from 1969 that did have a manual switch on the power supply to change the optimal load from 4 to 8 ohms.
Power amp specs are often un-trustworthy.  RMS power is a misnomer for mean power (the M in rms).
My question about your equipment was not an insult:  I noticed that you had not replied to my suggestions for simple measurements.  If you don't have access to or experience with such equipment, there's no further point in my suggesting measurement techniques.
« Last Edit: August 17, 2020, 06:48:48 pm by TimFox »
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #31 on: August 17, 2020, 07:04:13 pm »
To your quick question:  to change the ratio of voltage to current (i.e., the load resistance) without changing the DC supply and active devices, a transformer is needed to keep the same power.  A "step-up" transformer will give more volts and less amps into a higher resistance, with the same power (ignoring  transformer losses).  If "N" is the turns ratio, then V2 / V1  =  N  =  I1 / I2  (note the opposite order for V and I),  R2 / R1  = N2 , and P2 = P1.
Solid-state designs avoid transformers, and are optimized for one load impedance, although they will function over a reasonable range of load.  I remember a solid-state linear amplifier from 1969 that did have a manual switch on the power supply to change the optimal load from 4 to 8 ohms.
Power amp specs are often un-trustworthy.  RMS power is a misnomer for mean power (the M in rms).
My question about your equipment was not an insult:  I noticed that you had not replied to my suggestions for simple measurements.  If you don't have access to or experience with such equipment, there's no further point in my suggesting measurement techniques.

I didn't take it as an insult. The best way to do this is like you suggested, get out the equipment an just test it, done! (And, BTW, test it for THD at different levels too, and the type of distortion, and all the rest!)  I don't have that equipment nor the knowledge to run it.


So when doing calculations with a digital power supply, like the power brick included with a class D amp, you get the amps and voltage that the equation spits out and that's it?

If the power brick is 24v, and the amp needs 28V to get 50 watts per channel, how do you get 100 watts total RMS out of that amp using x2 3116D2 chips?

So how can we tell what the power of this amp is, generally, at 8 ohms, without actually measuring it--don't we have enough information to guesstimate it pretty closely?

Also, so it can put out 36 watts RMS x2 with two 3116D chips, it can do the same with one chip--so again, what's the benefit of two chips vs one chip at 8 ohms?
 

Offline TimFox

  • Super Contributor
  • ***
  • Posts: 9003
  • Country: us
  • Retired, now restoring antique test equipment
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #32 on: August 17, 2020, 07:30:44 pm »
If you know the voltage and maximum current of the power supply, and you understand the mode of operation of the active amplifier and its losses, then they determine the optimal load resistance and the power into it.  Usually, the power supply is treated as a constant voltage output with a maximum current capability, but the voltage will fall with higher current and excessive current could smoke the supply in a bad design.             
The details of that calculation are important, and you should consult the relevant textbooks.  I have used Class-D, but am not an expert on your chipset. 
At the optimal load, the peak current (of the sinusoidal waveform) into the load (maximum possible from the circuit) will require the maximum voltage that the amplifier can put out.  With a higher load resistance, the maximum peak voltage into the load requires less current and the power is lower.  With a lower load resistance, the maximum peak current requires a lower peak voltage and the power is lower.  Note the usage here of maximum (what the circuit can produce) and peak (what the load will do with the voltage or current supplied to it).  In any of these three cases, if you try to exceed either the maximum voltage or current (whichever is hit first as you increase the signal), the output waveform will show clipping.  Here, "peak" means the highest voltage (or current) in the waveform, half the "peak-to-peak" value.
What complicates all of this is that it is usually uneconomical (for home audio use) for the DC power supply to maintain its voltage for a long time at the maximum current.  Therefore, if you do a continuous measurement, the DC voltage probably "droops" after a short time and limits the power below that which could be delivered for a brief moment (as in normal music).
 
The following users thanked this post: DW1961

Online rsjsouza

  • Super Contributor
  • ***
  • Posts: 6107
  • Country: us
  • Eternally curious
    • Vbe - vídeo blog eletrônico
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #33 on: August 17, 2020, 09:58:23 pm »
"If I understood correctly your amplifier has two devices in PBTL (mono) mode and each is driving its own 2\$\Omega\$ loudspeaker. If so, figure 22 tells me that, at 24V, it can go as high as 150W each device, rendering your 24V/5A rather inadequate to supply this "beast". To get absolute maximum continuous power, you would need to have a minimum of 300W + efficiency losses. "

I dunno. That's why I'm trying to figure out if how the magic in these chips works.

The single 3116D2 chip runs in BTL mode, which tells me it is really 4 amps on one chip. In BTL mode, you need two chips running mono. However, the 3116D chip runs in stereo. So if it is running in BTL mode all of the time, and it is stereo, then it must have 4 amps on the chip. That's how they can run x2 3116D chips in mono and get stereo in the amp I have, I assume. Otherwise, if it were just two amp chips, they would be only getting a mono amp out of it. So the way I understand it, to get stereo from a BTL system, you need 4 amps.

So the amp I have must really have 8 amps on it (x2 3116D2 chips @ 4 amplifiers each), but they are running the 3116D in mono mode, so they need another one to do stereo.

How they are connecting them together, I have no idea! PBTL? I just dunno.
Section 7.4.1 shows the PBTL (mono) connection.

It is important to keep the nomenclature identical to the datasheet - in this case the authoritative source of information. You are correct the device internally has four amplifiers connected two-by-two in BTL that could be bridged yet again to form a single mono amplifier, to which the datasheet calls it "PBTL mode".

"So, why are you talking about 2\$\Omega\$? Sorry, I really don't get exactly the circling around. "

To get 100 watts RMS from one 3116D, it has to run at 2 Ohms in mono.
Nope. As Kai's reply on e2e also clarifies, to get 100W RMS from one 3116D you can run it in BTL (stereo) at 4 ohms, which yields 50W per channel or 100W total, or in PBTL (mono) at 2 ohms, which yields 100W on a single channel. According to figures 14 and 22, you can get all this power using "only" 19V.

Keep going up in voltage and you will get up to 170W from a single 3116D, either in BTL or PBTL modes.

"24V * 5A = 120W. Where did the 72W come from? "
Using this calculator: https://www.rapidtables.com/calc/electric/watt-volt-amp-calculator.html
Add resistance at 8 ohms: 8 ohms/3 amps/24v/72 watts. You could increase current, but then you would need to increase voltage too. I guess Ohms law?
Well, a resistor would consume 72W, which is equal to 242/8, but that is where you need to consider the amplifier efficiency. The datasheet tells you that 42W per channel is the maximum under these load and supply conditions. As TimFox mentioned, the amplifier will draw whatever it needs from the power supply - in other words, the power supply will passively provide whatever is within its limits, compensating with a reduction in voltage in case the amplifier consumes more than it will provide (thus clipping the output voltage).
Vbe - vídeo blog eletrônico http://videos.vbeletronico.com

Oh, the "whys" of the datasheets... The information is there not to be an axiomatic truth, but instead each speck of data must be slowly inhaled while carefully performing a deep search inside oneself to find the true metaphysical sense...
 
The following users thanked this post: DW1961

Offline cv007

  • Frequent Contributor
  • **
  • Posts: 862
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #34 on: August 18, 2020, 12:57:06 am »
>What is the benefit of running two of these when the power stays the same, and one amp chip is capable of putting out the same wattage using the same driver for both chips?


You also have a set of mosfets that have to dissipate power (8x120 milliohm 'resistors', 4 active at a time). If a single ic is 'heat happy' putting out 50Wx2 (BTL, 100W total), then its probably also happy putting out 100Wx1 (PBTL, 100W total). Since there is a voltage limit, the only way to get to the 100W is by decreasing the impedance. The PBTL mode is simply a way to share the heat. Since there is no perfect mosfet, I imagine heat is the monster that lurks everywhere in the datasheet.

If you are using 8ohm speakers, you cannot take advantage of the possible extra power (because of voltage limit), but you still get the benefit of sharing the heat among 2 ic's. That may mean you do not need a heatsink, or could mean the wimpy heatsinks used in an enclosed little case will not cause thermal shutdown.

You need to look at any seller of audio gear the same way you look at laundry detergent that says '2x the cleaning power'. Its a waste of time trying to sort out the claims being made. Just look beyond the sales pitch, and look at the ingredients to figure out what you have.
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #35 on: August 18, 2020, 01:43:54 am »

Nope. As Kai's reply on e2e also clarifies, to get 100W RMS from one 3116D you can run it in BTL (stereo) at 4 ohms, which yields 50W per channel or 100W total, or in PBTL (mono) at 2 ohms, which yields 100W on a single channel. According to figures 14 and 22, you can get all this power using "only" 19V.

Keep going up in voltage and you will get up to 170W from a single 3116D, either in BTL or PBTL modes.

"24V * 5A = 120W. Where did the 72W come from? "
Using this calculator: https://www.rapidtables.com/calc/electric/watt-volt-amp-calculator.html
Add resistance at 8 ohms: 8 ohms/3 amps/24v/72 watts. You could increase current, but then you would need to increase voltage too. I guess Ohms law?
Well, a resistor would consume 72W, which is equal to 242/8, but that is where you need to consider the amplifier efficiency. The datasheet tells you that 42W per channel is the maximum under these load and supply conditions. As TimFox mentioned, the amplifier will draw whatever it needs from the power supply - in other words, the power supply will passively provide whatever is within its limits, compensating with a reduction in voltage in case the amplifier consumes more than it will provide (thus clipping the output voltage).

"Nope. As Kai's reply on e2e also clarifies, to get 100W RMS from one 3116D you can run it in BTL (stereo) at 4 ohms, which yields 50W per channel or 100W total, or in PBTL (mono) at 2 ohms, which yields 100W on a single channel. According to figures 14 and 22, you can get all this power using "only" 19V. "

Yep, I failed to correct that statement last night. Sure, it says right there on the spec sheet a single 3116D amp will do 100 watts total at 4 ohms and 21 V. And, 100 watts mono at 2 ohms.

"Keep going up in voltage and you will get up to 170W from a single 3116D, either in BTL or PBTL modes. "

Ok, yes, from one chip running at 4 ohms with 26V and 6.5 amps = 170 watts total.


And we know that a single 3116D amp chip can put out 100 watts RMS at 4 ohms @ 21V (top of spec page).

Using the chart on page ten for a single 3116D and 8 ohms and 26V, which is maximum voltage the chip is rated for:
Approx. 85 watts total RMS at 26V @3.3A.


My question is:

Using two of these amp chips in PBTL mode for stereo, and leaving out efficiency factors of the amp package they come in, how many total watts RMS is it going to put out at 8 ohms using a 24V/5A PSU*? And, to my point, is to going to be anymore than using the same PSU with a single 3116D chip at 8 ohms?

*Because that's what comes with the amp package.
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #36 on: August 18, 2020, 02:06:48 am »
>What is the benefit of running two of these when the power stays the same, and one amp chip is capable of putting out the same wattage using the same driver for both chips?


You also have a set of mosfets that have to dissipate power (8x120 milliohm 'resistors', 4 active at a time). If a single ic is 'heat happy' putting out 50Wx2 (BTL, 100W total), then its probably also happy putting out 100Wx1 (PBTL, 100W total). Since there is a voltage limit, the only way to get to the 100W is by decreasing the impedance. The PBTL mode is simply a way to share the heat. Since there is no perfect mosfet, I imagine heat is the monster that lurks everywhere in the datasheet.

If you are using 8ohm speakers, you cannot take advantage of the possible extra power (because of voltage limit), but you still get the benefit of sharing the heat among 2 ic's. That may mean you do not need a heatsink, or could mean the wimpy heatsinks used in an enclosed little case will not cause thermal shutdown.

You need to look at any seller of audio gear the same way you look at laundry detergent that says '2x the cleaning power'. Its a waste of time trying to sort out the claims being made. Just look beyond the sales pitch, and look at the ingredients to figure out what you have.

Yeah agree with all of that, for sure. That's why I originally started the thread.
 

Online rsjsouza

  • Super Contributor
  • ***
  • Posts: 6107
  • Country: us
  • Eternally curious
    • Vbe - vídeo blog eletrônico
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #37 on: August 18, 2020, 02:36:07 am »
My question is:

Using two of these a,mp chips in PBTL mode for stereo, and leaving out efficiency factors of the amp package they come in, how many total watts RMS is it going to put out at 8 ohms using a 24V/5A PSU*? And, to my point, is to going to be anymore than using the same PSU with a single 3116D chip at 8 ohms?

*Because that's what comes with the amp package.
TI did not seem to have characterized the device under these conditions, thus any conclusion is pure extrapolation based on the data.

In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL, thus yielding 42W * 2 or 84W (THD+N = 10%) - figure 13 shows 42W in BTL (stereo) mode when driving 8 ohms at 24V.

Given the power supply can put out 120W, it is enough for a single 3116D operating in PBTL at 8 ohms at maximum output continuous power.

Putting two 3116Ds in PBTL would require at least 168W + efficiency losses (about 87% or 88% according to the various graphs) - about 200W with safe margin or around 24V / 8.5A.

Hopefully I didn't mess up my calculations.
Vbe - vídeo blog eletrônico http://videos.vbeletronico.com

Oh, the "whys" of the datasheets... The information is there not to be an axiomatic truth, but instead each speck of data must be slowly inhaled while carefully performing a deep search inside oneself to find the true metaphysical sense...
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #38 on: August 18, 2020, 06:24:20 am »
My question is:

Using two of these a,mp chips in PBTL mode for stereo, and leaving out efficiency factors of the amp package they come in, how many total watts RMS is it going to put out at 8 ohms using a 24V/5A PSU*? And, to my point, is to going to be anymore than using the same PSU with a single 3116D chip at 8 ohms?

*Because that's what comes with the amp package.
TI did not seem to have characterized the device under these conditions, thus any conclusion is pure extrapolation based on the data.

In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL, thus yielding 42W * 2 or 84W (THD+N = 10%) - figure 13 shows 42W in BTL (stereo) mode when driving 8 ohms at 24V.

Given the power supply can put out 120W, it is enough for a single 3116D operating in PBTL at 8 ohms at maximum output continuous power.

Putting two 3116Ds in PBTL would require at least 168W + efficiency losses (about 87% or 88% according to the various graphs) - about 200W with safe margin or around 24V / 8.5A.

Hopefully I didn't mess up my calculations.

Regardless of your calculations, I think you are saying the same thing I was thinking when I started this thread: How do you get any more power out of two chips vs one chip when the power supply only puts out 120 watts @ 0hms or 72 watts at 24v/8 ohms?

"Given the power supply can put out 120W, it is enough for a single 3116D operating in PBTL at 8 ohms at maximum output continuous power."

It cannot put that much power out when you add resistance: 72 watts @ 8 ohms and you run up against the PSU's voltage max, which is 24v. (If I am understanding this now.)

It just seems like they can do everything with one chip vs two, especially since the amp package has a sub out for an active sub. I mean at 21V and just under 5 amps, you can get what they say at 4 ohms: 100 watts. IF you had ONE chip in the same amp package with a sub out for an active sub, you would still have the same power to play with given you are usually not going any lower than  4 ohms.

But when they say you get 100 watts at 2 ohms, sure, if you have a PSU that puts out 7 amps at 14v.

Again, the PSU we're dealing with is 24v/5A. That's all we get. It would be different is you had a 24V 10A power supply, then yeah, you would get a LOT more benefit form using 2 chips vs one becasue you would run up against voltage  restrictions using one chip and would need two of them to put out more power.

For instance, using a 26V power supply with a single chip, you can get max RMS at 8 ohms (according to the graph) and 3.3 amps of 86 watts. If you want more power at 8 ohms, then you need to use two of them. And, you would need to increase the PSU voltage (not necessarily the amperage) past 26V, which I assume could be regulated to each chip to not exceed 26V. Then, yeah, I can see the benefit of using two of them, but not with the same power supply that only puts out 24V.

So the question remains, why build a dual 3116D amp with a 24V/5A power supply?

 

Offline cv007

  • Frequent Contributor
  • **
  • Posts: 862
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #39 on: August 18, 2020, 06:56:29 am »
>In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL

You are missing the fact that the PBTL graph is using 2 Ohms. All the graphs essentially end up the same (BTL 8,4,PBTL 2), each is just changing the load impedance to increase the power. PBTL is already BTL, its just parallel to share the heat.

If you have 24v and 8 Ohms, it makes no difference if you run in BTL or PBTL as your total output will be the same. What you get, as I have said previously, is the ability to direct all the power into a single low impedance speaker if wanted. With 8 ohms and PBTL, all you get is less heat for each ic, which is a good enough reason and could be the main reason they run them in PBTL.

 
The following users thanked this post: tooki, DW1961

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #40 on: August 18, 2020, 05:48:27 pm »
>In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL

You are missing the fact that the PBTL graph is using 2 Ohms. All the graphs essentially end up the same (BTL 8,4,PBTL 2), each is just changing the load impedance to increase the power. PBTL is already BTL, its just parallel to share the heat.

If you have 24v and 8 Ohms, it makes no difference if you run in BTL or PBTL as your total output will be the same. What you get, as I have said previously, is the ability to direct all the power into a single low impedance speaker if wanted. With 8 ohms and PBTL, all you get is less heat for each ic, which is a good enough reason and could be the main reason they run them in PBTL.

Spot on for resistance. The only way an amp "makes more power" is either getting more voltage from the supply or lowering resistance so more current can flow to it. At least that's how I am understanding it currently.

"If you have 24v and 8 Ohms, it makes no difference if you run in BTL or PBTL as your total output will be the same." Clear, and thank you.

What I have been saying all  along in this thread in one way or another, which is why I created the thread. I just didn't see the power benefit of using two chips vs one given the same voltage restrictions.

"all you get is less heat for each ic, which is a good enough reason and could be the main reason they run them in PBTL." ---Yo!, thanks.

However, the single amp chip says it can run full out without any problems heat wise. If they really wanted to reduce chip heat, it seems like they cold do it a lot cheaper buy using a bigger heat sink or active cooling. Fans and heat sinks are a lot cheaper than the 3116D chip, I would assume, and less engineering too.

The seller on Amazon wrote me back and said the amp package they offer with x2 3116D chips is 24V and anything higher will burn it. He also said that using the included power supply at 24V, you're gonna get 70 watts continuous power at 8 ohms. That's the same as a single chip, according to the graph on the spec sheet we've been referring to.
 

I keep asking them why use 2 chips then? We'll see if they ever answer.
 

Offline tooki

  • Super Contributor
  • ***
  • Posts: 13157
  • Country: ch
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #41 on: August 18, 2020, 06:51:32 pm »
You’re not going to get a reply from an engineer. Expecting a reply like that is an exercise in frustration.
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #42 on: August 18, 2020, 07:21:44 pm »
You’re not going to get a reply from an engineer. Expecting a reply like that is an exercise in frustration.


LOL, don't I know it! Well, when I review that product on Amazon and explain to people they are actually getting 70w atts RMS at 8 ohms, at least there's that. The seller actually admitted that in the public comments section, though.

I tried one of these dual 316D amps before and it seemed like it put out more power, not at a static 8 phms, but just turning up the dial listening to music, it seems to have a lot more punch. No way to tell.

The problem I had is that it was noisy when turning up the volume knob and with no music playing. It had a lot of hiss. My single amp 3116D unit does not have that. It's really quite. It was the same vendor's product.

This new unit I'll test as well as I can tonight for hiss or any other artifacts. It's also a Bluetooth 5.1 receiver and as stated before, has a passive sub out for an active sub connection. I think this is where any of these 3116D chips could really shine, when you use the 3116D power to power mids and highs only, and use an active sub for the rest

However, you could do the same thing with a single chip and add a passive sub out to it.
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #43 on: August 19, 2020, 06:03:07 am »
I had an epiphany!

Someone would have to do the calculation, maybe I will.

But! I think I have figured out one  benefit  from using (2) 3116D chips.

First, I just want to make sure my calcs are right about this included PSU. The included PSU for this unit is 24V/4.5A. That means the most power you will get with that PSU at 2 ohms is 40 watts total--lol? You need 10 A to get 200 watts at 2 ohms. Am I seeing that correctly?

Anyway, here's my idea. Let's keep the voltage at 24V even though the specs say you can use 26V, and let's allow amps to be whatever they are. (Did anyone actually see an amp rating for these chips? I'm assuming it's 7 A because it needs 7 amps at 2 ohms in mono to make 100 watts.)

You could wire (4) 4 ohms speakers in parallel and get 2 ohms or you could use (4) 8 ohms speakers at 4 ohms. Using a single chip, the most power you will get at 24V and 4 ohms is 100 watts.

If you wire up 4-8 ohms speakers you get get a 4 ohm speaker system at 100 watts shared, or 25 per channel.

FOUR SPEAKER SYSTEM
4 0hms
5 A
20 V
= 100 watts.
25 watts per speaker

TWO  SPEAKER SYSTEM
8 ohms
3A
24V
= 72 watts
36 watts per speaker.

You lose 30% of your power, but you gain 100% more speaker. Just off the top of my head, you're going to make a lot more volume with the 4 speaker setup at 4 ohms than using 2 at 8 ohms (depending on speaker sensitivity).

Am I far enough down the rabbit hole yet?
 

Online rsjsouza

  • Super Contributor
  • ***
  • Posts: 6107
  • Country: us
  • Eternally curious
    • Vbe - vídeo blog eletrônico
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #44 on: August 19, 2020, 03:25:48 pm »
>In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL

You are missing the fact that the PBTL graph is using 2 Ohms. All the graphs essentially end up the same (BTL 8,4,PBTL 2), each is just changing the load impedance to increase the power. PBTL is already BTL, its just parallel to share the heat.
I am not missing anything. I am just using the output power characterized in Figure 13 (42-ish W for 24V in 8 ohms) and doubling it for PBTL, just like I did when extrapolating figure 14 to PBTL in 4 ohms (which coincidentally matches 2 ohms in PBTL).

Oh well... Take it as it is. I would go with the manufacturer's specs and characterization curves instead of assuming that all is equivalent and equal.

Good luck with your amp, DW1961. Peace out.
Vbe - vídeo blog eletrônico http://videos.vbeletronico.com

Oh, the "whys" of the datasheets... The information is there not to be an axiomatic truth, but instead each speck of data must be slowly inhaled while carefully performing a deep search inside oneself to find the true metaphysical sense...
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #45 on: August 19, 2020, 04:41:38 pm »
>In this case, I would assume the relationship between figures 14 and 22 could also be applied to figure 13, that is: at 24V, power at 8 ohms in PBTL would be double when compared to BTL

You are missing the fact that the PBTL graph is using 2 Ohms. All the graphs essentially end up the same (BTL 8,4,PBTL 2), each is just changing the load impedance to increase the power. PBTL is already BTL, its just parallel to share the heat.
I am not missing anything. I am just using the output power characterized in Figure 13 (42-ish W for 24V in 8 ohms) and doubling it for PBTL, just like I did when extrapolating figure 14 to PBTL in 4 ohms (which coincidentally matches 2 ohms in PBTL).

Oh well... Take it as it is. I would go with the manufacturer's specs and characterization curves instead of assuming that all is equivalent and equal.

Good luck with your amp, DW1961. Peace out.

RJ, you do realize that you were replying to cv007, right, and not me? Just wanted to make sure. :)
 

Online rsjsouza

  • Super Contributor
  • ***
  • Posts: 6107
  • Country: us
  • Eternally curious
    • Vbe - vídeo blog eletrônico
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #46 on: August 20, 2020, 12:59:56 am »
RJ, you do realize that you were replying to cv007, right, and not me? Just wanted to make sure. :)

You are alright, DW. Everybody is alright, really; sorry for coming in too hard. New job, stupid quarantine and first day of virtual school with my twins takes a toll on one's patience.

I hope the specs are clearer for you. Cheers!
Vbe - vídeo blog eletrônico http://videos.vbeletronico.com

Oh, the "whys" of the datasheets... The information is there not to be an axiomatic truth, but instead each speck of data must be slowly inhaled while carefully performing a deep search inside oneself to find the true metaphysical sense...
 

Offline DW1961Topic starter

  • Frequent Contributor
  • **
  • Posts: 753
  • Country: us
Re: Texas Instruments 3116D (ClassD) amp chip
« Reply #47 on: August 29, 2020, 04:55:26 am »
RJ, you do realize that you were replying to cv007, right, and not me? Just wanted to make sure. :)

You are alright, DW. Everybody is alright, really; sorry for coming in too hard. New job, stupid quarantine and first day of virtual school with my twins takes a toll on one's patience.

I hope the specs are clearer for you. Cheers!

It's all good. Yeah, I know now that there is no benefit from running 2 of those with their power supply. It's just all hype. I'd like to see some company do it right and use a 12A power supply to power both those chips to full power, though. Then you could get 50 watts per channel at 8 ohms, and 100 watts per channel at 4 ohms. Now we're talking some serious power.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf