I'd be grateful if someone would take a look at the data sheet for the PAM8403: http://www.allelectronics.com/mas_assets/theme/allelectronics/spec/AMP-8403.pdf
I would like some help with a decision to buy either a 4 or 8 ohm versions of a 3" full range driver with a rated sensitivity of 91db/1w/1m. I am keen to find out if the driver choice will affect battery life or max volume before distortion.
Given both 4 and 8 ohm versions are 91db/1w/1m is it safe to say the power used will be the same?
The PAM datasheet THD+N graphs indicate that for a 5v supply an 8 ohm driver can be supplied with 1.6watts of power before THD+N>1%. Given the same parameters and a 4ohm driver the amplifier can supply 2.6watts of power before THD+N>1%.
Given this, would the 4ohm version be the better choice? are there other factors to consider?
I would like some help with a decision to buy either a 4 or 8 ohm versions of a 3" full range driver with a rated sensitivity of 91db/1w/1m. I am keen to find out if the driver choice will affect battery life or max volume before distortion.
Given both 4 and 8 ohm versions are 91db/1w/1m is it safe to say the power used will be the same?
The PAM datasheet THD+N graphs indicate that for a 5v supply an 8 ohm driver can be supplied with 1.6watts of power before THD+N>1%. Given the same parameters and a 4ohm driver the amplifier can supply 2.6watts of power before THD+N>1%.
Given this, would the 4ohm version be the better choice? are there other factors to consider?
The PAM8403 is about 5% more efficient when driving an 8 ohm load versus a 4 ohm load from about 0.25W output and up. I would use an 8 ohm load unless you need the extra power for the extra few dB of output with a 4 ohm load.
Edit: what 3" full range driver has a sensitivity of 91dB @ 1W/m? Unless the driver is horn loaded, that sounds like it's a peak specification, not a reasonable average of the frequency response.
Edit: what 3" full range driver has a sensitivity of 91dB @ 1W/m? Unless the driver is horn loaded, that sounds like it's a peak specification, not a reasonable average of the frequency response.
Last edited:
Hi Brian,
Thanks for your response and pointing to the increased efficiency of the 8 ohm, looking at the spec sheet it is 87% vs 83%.
I have ordered the 8 ohm versions of the driver. The driver is the Faital pro 3FE22 and the SPL is 91db/1w/1m. I have used it's Ferrite sibling the 3FE25 with the same sensitivity for a MLTL, and in my humble opinion it's a cracking driver. I have also used it's bigger brother the 4FE22, bigger sound but not as refined in my opinion.
As far as the THD+N issue, my thinking was off. Although with a 4 ohm load, the amp can supply more wattage with a <1%THD+N, for a given volume level the current would double opposed to an 8ohm load. Given a constant voltage output, the power from the amp would double and so the THD+N are similar. I also assume the power drawer from the battery pack would also increase for the 4 ohm load.
Please intervene if my thinking is way off here!
Thanks for your response and pointing to the increased efficiency of the 8 ohm, looking at the spec sheet it is 87% vs 83%.
I have ordered the 8 ohm versions of the driver. The driver is the Faital pro 3FE22 and the SPL is 91db/1w/1m. I have used it's Ferrite sibling the 3FE25 with the same sensitivity for a MLTL, and in my humble opinion it's a cracking driver. I have also used it's bigger brother the 4FE22, bigger sound but not as refined in my opinion.
As far as the THD+N issue, my thinking was off. Although with a 4 ohm load, the amp can supply more wattage with a <1%THD+N, for a given volume level the current would double opposed to an 8ohm load. Given a constant voltage output, the power from the amp would double and so the THD+N are similar. I also assume the power drawer from the battery pack would also increase for the 4 ohm load.
Please intervene if my thinking is way off here!
When entering the TS parameters in Hornresp and looking at the directivity response on axis in an infinite baffle the maximum driver output in the midrange is around 87dB, which is more in line with what a high sensitivity 3 inch driver is probably capable of. It would be interesting to take some actual measurements of the 3FE22 to see if Faital Pro's claim of 91dB at 1W/m is actually true.
Assuming the IC can provide a maximum of 5V output (bridged) and twice the current to a 4 ohm load, the maximum theoretical RMS power for a sine wave signal into 4/8 ohm loads would be 3.13/1.56W. Obviously it can't do that with any reasonable amount of THD, so you need to consider the maximum level you'll want to use this at. For 1W output with perfectly resistive loads it would be 2Vrms, 0.5Arms with 4 ohms and 2.83Vrms, 0.35Arms with 8 ohms. The efficiency is higher with an 8 ohm load because the output current is lower.
Pout=Pin*Efficiency or Vout*Iout=Vin*Iin*Efficiency
Assuming the IC can provide a maximum of 5V output (bridged) and twice the current to a 4 ohm load, the maximum theoretical RMS power for a sine wave signal into 4/8 ohm loads would be 3.13/1.56W. Obviously it can't do that with any reasonable amount of THD, so you need to consider the maximum level you'll want to use this at. For 1W output with perfectly resistive loads it would be 2Vrms, 0.5Arms with 4 ohms and 2.83Vrms, 0.35Arms with 8 ohms. The efficiency is higher with an 8 ohm load because the output current is lower.
Pout=Pin*Efficiency or Vout*Iout=Vin*Iin*Efficiency
Last edited:
Arh yes. I was working on the basis both speakers were fed 2.83V to achieve a given volume, in which case the current would be approx 7amps for the 4ohm load and 0.35amps for the 8ohm load. Given the specs are for 1w, the 4ohm load as you stated will be 2V.
The modeling I have done with this driver in MK's MLTL worksheet has the flat section of the frequency range hitting 88db. The manufacturers graph shows 90db+ over 500Hz+ frequencies.
Any idea of the setup to measure driver sensitivity?
The modeling I have done with this driver in MK's MLTL worksheet has the flat section of the frequency range hitting 88db. The manufacturers graph shows 90db+ over 500Hz+ frequencies.
Any idea of the setup to measure driver sensitivity?
You probably won't want to push the IC much past 1% THD, which is about 2.5W into 4 ohms or 1.5W into 8 ohms, or it will start sounding pretty bad. At 10% THD the output is definitely clipped, so don't even use those inflated output power specs. Assuming the drivers are actually 88dB at 1W/m that'll get you a maximum SPL at 1m of slightly more than 91dB with the 4 ohm driver or slightly more than 89dB with the 8 ohm driver. Not a significant difference either way.
You'll need either a SPL meter or a mic and REW (or similar software) to measure the sensitivity. A UMIK-1 is $85 from Parts Express and REW is free as long as you have a computer to run it on. There are less expensive options for mics as well, so it's not prohibitively expensive to get yourself a decent measurement setup.
You'll need either a SPL meter or a mic and REW (or similar software) to measure the sensitivity. A UMIK-1 is $85 from Parts Express and REW is free as long as you have a computer to run it on. There are less expensive options for mics as well, so it's not prohibitively expensive to get yourself a decent measurement setup.
Last edited:
I have just received the 8ohm versions, so I'm up for testing their sensitivity. I have both REW and a umik-1, although I'm just cutting my teeth with REW. Can I check through the test method:
1) Place the driver in a baffle min 115mm (1000Hz/115)
2) Generate a 1kHz sine wave with REW.
3) Adjust the amp volume with a multimeter on the speaker terminals, set volume so voltage peaks at 4v or 2.83RMS.
4) Setup Mic at 1m and read REW spl meter.
1) Place the driver in a baffle min 115mm (1000Hz/115)
2) Generate a 1kHz sine wave with REW.
3) Adjust the amp volume with a multimeter on the speaker terminals, set volume so voltage peaks at 4v or 2.83RMS.
4) Setup Mic at 1m and read REW spl meter.
At 1kHz the wavelength is 344/1000=0.344m. I don't know what the dimensions are for a properly sized IEC baffle, but I think they're fairly large. Whatever baffle size you use, don't mount the driver directly in the center, it should be offset an unequal amount in both x and y dimensions. Alternatively, you could just measure it in a sealed box similar to what you'll actually use, calculate what the baffle step loss should be, and estimate what the response would have been in a much larger baffle.
The rest of your test method looks good.
The rest of your test method looks good.
Arh yes I see going wavelength makes sense. The 115mm figure I got from John Murphy's book. But this is the centre at which the 6db step occurs and so at 115mm I'd add 3db to the spl measured.
For a closed box with a baffle width of 200mm (less than the wavelength of the test frequency) how would one calculate the diffraction loss? Assuming a 1kHz test tone.
For a closed box with a baffle width of 200mm (less than the wavelength of the test frequency) how would one calculate the diffraction loss? Assuming a 1kHz test tone.
Last edited:
All the info is on John Murphy's web page: https://www.trueaudio.com/st_diff1.htm
It's really more of an estimation because of all the ripples that will be in the frequency response.
It's really more of an estimation because of all the ripples that will be in the frequency response.
Yes I read this, which is where I got the 115mm wide baffle from. At this baffle width I think I'd add 3db to the measure spl. I was just curious to see if there is a way of determining the gradient of the 6db baffle step, but perhaps this is determined from too many factors to model. It seems sensible to me to go with a baffle width>wavelength and then there is zero diffraction. I'll report back one I have measured.
I think the roll off due to baffle width takes an octave, as it changes from half space loading to full space loading.
That might be between the F-3dB points on the frequency response curve. It will result in quite a slow rate of slope/octave, probably less than 3dB/octave at the steepest part of the slope.
That might be between the F-3dB points on the frequency response curve. It will result in quite a slow rate of slope/octave, probably less than 3dB/octave at the steepest part of the slope.
I was coming to the same conclusion Andrew. After reading that an obstacle is audibly significant when it equates to 0.5 wavelength. And logically a baffle = full wavelength would be half space, the octave seems correct. But many published baffle step responses seem spread over 2+octaves. I'm still baffled. Aha.
I just tested the driver at 300Hz (Volt Meter rated <400Hz) , baffle width 500mm, sine wave, 2.75V RMS across the speaker terminals, calibrated mic pickup positioned 0 deg 1m in front of baffle and the reading was 93.2dB. Higher than the Faital Pro response graph which shows approx 88bB at 300Hz. Not sure what's up with the result I generated, but it looks like there must be an error. When I set the amp gain so there was 2.75V RMS (closest I could get to 2.83V as my amp gain is in 0.5dB increments) across the speaker terminals, and then lowered the frequency the voltage increased, and when the frequency increased the voltage decreased. Is this normal? Any ideas for tracking down the error?
Last edited:
Agreed, the voltmeter does find AC sine waves tricky. One quirk I discovered with my meter is that if the frequency changes during a read the value fluctuates. If the signal is stopped and then a fresh read is taken it seems to be consistent. The voltmeter spec is to read AC voltage 40-400Hz with 0.8% accuracy at the level tested. I got tightly groups reads from 50-400Hz, and then generated a sweep in REW from 90Hz-20kHz and recorded the attached. When I ironed out the result with 1/1 smoothing the average was approx 90dB. Although the response is all over the place (probably contributed to by room and furniture effects), the average sensitivity isn't far off the manufacturers spec.
Attachments
The 3 inch driver should be beaming above about 5k, so it's easier to compare the response in that range. Assuming you were driving it at 1W (seems like you should have been pretty close) your measurements are slightly lower at 3k (87-88dB versus 90dB) but rise up to similar levels at 10k (94-95dB) before really rolling off from 13k-20k (is your source or amp causing that?). Without using the same baffle you will obviously get different results at lower frequencies, but it still does look like a very sensitive driver. Should give you some decent SPL from the small drivers and the limited power output of the PAM8403.
You can't usefully test a loudspeaker, in a room, without massive attention to room return and usually smoothing. The irregularities are eye-opening (since this is the same path as to your ear).
I would take the claimed "sensitivity" as sorta-right, more right than you can confirm at home.
I would take the claimed frequency response with massive doses of salt. A 3" at 91db/1w/1m can have a flat Power Response from 2KHz down to around 250Hz sealed box, ~~120Hz with a precisely designed vent box. Above 2KHz it will, as BWRX says, be beamy (declining off-axis, but flattish on-axis). I'm not sure it is "easier to compare" in the beam zone, but it will be less and less room response mucking your measurements. Unless deliberately limited, I'd expect flat or slightly rising to 15KHz.
On a half-meter baffle I'd expect disturbance and bass-cut around 340Hz (half wave) which is seen in your plot.
I would take the claimed "sensitivity" as sorta-right, more right than you can confirm at home.
I would take the claimed frequency response with massive doses of salt. A 3" at 91db/1w/1m can have a flat Power Response from 2KHz down to around 250Hz sealed box, ~~120Hz with a precisely designed vent box. Above 2KHz it will, as BWRX says, be beamy (declining off-axis, but flattish on-axis). I'm not sure it is "easier to compare" in the beam zone, but it will be less and less room response mucking your measurements. Unless deliberately limited, I'd expect flat or slightly rising to 15KHz.
On a half-meter baffle I'd expect disturbance and bass-cut around 340Hz (half wave) which is seen in your plot.
- Status
- Not open for further replies.