In this post I test Altec 288-8G Compression Driver with Altec 805B Multicell.
I borrowed the horn from my brother.
Specification
Compression Driver
Frequency Response
I measured the horn with a 2m mic distance. Above 1kHz my measurement is gated to 4ms using 1/24 octave smoothing. Below 1kHz the repones is ungated and 1/3 octave smoothing applied. Generally we see a reasonably smooth response with a midrange emphasis centered at about 2kHz. The horn loads strongly down to 400Hz with very little first order reflection near the the horn's cutoff frequency which is published as 500Hz. The diaphragm is breakup free until 13.5kHz.
Below is the horizontal off-axis colored polar map using the same 2m mic distance.
Characteristics
Time Domain
Conducted in the near field using a 20cm mic distance.
The burst decay shows some stored energy in the treble region.
Distortion
I measured distortion with the 20cm mic distance. SPL level was set at 85dB and 95dB referenced at 1m. I set a fixed resistor L-pad using a 20w 7.5ohm and 1.5ohm (R1 & R2) which provides about -15dB attenuation. This brought the amplifier up out of its noise floor (Hypex FA500). Above 1kHz we see H3 and H4 at very low levels while H2 dominates. H3 for the 95dB signal is only 0.015% for the 1kHz region.
Multitone is shown below for the same 85dB and 95dB test signal levels.
The 85dB test signal shows 70dB dynamic range for the 1.5kHz region reducing to 60dB for the 10kHz region.
The 95dB test signal is unchanged for the midrange but we see 10kHz reduce to 50dB dynamic range.
Contour Circuit
The horn has a strong midrange presence along with some coloration near the horn's cutoff frequency. So I decided to develop a basic high pass filter along with a contour circuit. I settled on a LR3 high pass filter to eliminate the coloration in the 500Hz region. I then implemented an parallel LCR contour to bring down the 2kHz region. The raw response is shown in red, the blue response is with the LCR, and the green response is with the high pass filter.
Subjective Listening
With contour circuit No.2567 in place, I've summarized my listening impressions.
I borrowed the horn from my brother.
Specification
Compression Driver
- 2.80" Diameter Voice Coil (71mm)
- Aluminum diaphragm
- 7.1" outside diameter for compression driver body
- 29.3 lbs compression driver weight (13.3 Kg)
- Alnico magnet
- 40 x 80 coverage pattern
- 500Hz Cutoff frequency (Fc)
- 62cm wide x 34cm tall x 55cm deep
Frequency Response
I measured the horn with a 2m mic distance. Above 1kHz my measurement is gated to 4ms using 1/24 octave smoothing. Below 1kHz the repones is ungated and 1/3 octave smoothing applied. Generally we see a reasonably smooth response with a midrange emphasis centered at about 2kHz. The horn loads strongly down to 400Hz with very little first order reflection near the the horn's cutoff frequency which is published as 500Hz. The diaphragm is breakup free until 13.5kHz.
Below is the horizontal off-axis colored polar map using the same 2m mic distance.
Characteristics
- 90 degree coverage (500Hz-1.5kHz)
- Widens to 120 degrees (1.5kHz-17kHz)
- High frequency output drops when measuring directly in front of each cell
Time Domain
Conducted in the near field using a 20cm mic distance.
The burst decay shows some stored energy in the treble region.
Distortion
I measured distortion with the 20cm mic distance. SPL level was set at 85dB and 95dB referenced at 1m. I set a fixed resistor L-pad using a 20w 7.5ohm and 1.5ohm (R1 & R2) which provides about -15dB attenuation. This brought the amplifier up out of its noise floor (Hypex FA500). Above 1kHz we see H3 and H4 at very low levels while H2 dominates. H3 for the 95dB signal is only 0.015% for the 1kHz region.
Multitone is shown below for the same 85dB and 95dB test signal levels.
The 85dB test signal shows 70dB dynamic range for the 1.5kHz region reducing to 60dB for the 10kHz region.
The 95dB test signal is unchanged for the midrange but we see 10kHz reduce to 50dB dynamic range.
Contour Circuit
The horn has a strong midrange presence along with some coloration near the horn's cutoff frequency. So I decided to develop a basic high pass filter along with a contour circuit. I settled on a LR3 high pass filter to eliminate the coloration in the 500Hz region. I then implemented an parallel LCR contour to bring down the 2kHz region. The raw response is shown in red, the blue response is with the LCR, and the green response is with the high pass filter.
Subjective Listening
With contour circuit No.2567 in place, I've summarized my listening impressions.
- Spatial and enveloping treble due to 120 degree wide coverage
- Despite the class D amplifier, the sound had some typical H2 smoothness.
- Missing upper treble detail
- Some horn coloration across its bandwidth
- Strong presence on vocals (voice modulation easy to follow)
- Some masking of soundstage depth
@Joseph Crowe
Thanks for the data, all of it is much appreciated!
Question: Are you going to lash that 288-8G up to one of your ES BiRadials ( such as the ES 450 ) for a measurement session? I would love to see the results ( as I imagine would others ).
Also, I'm curious about what the impedance curve for the 288G/805b combo looks like ( could you post a pic of it please?? ).
Does it resemble the following ( iow; having a well defined peak of around 300hz )?? ( this pic lifted from Altec's own TL-231 technical letter (pdf) ).
Compare that to this earlier incarnation of the 288. This is a "C" model ( with a 16 ohm diaphragm ) in this case on a 329a Radial .
I do wonder if that impedance peak seen at 500hz is//was in fact accurate for that older diaphragm ( or possibly a mistake in the setup in the equipment for the combining of the twin traces on the B&K graph paper ).
Here's a Bill Woods ( RIP ) trace that he did of a 288C on a 1003b multicell.
- Followed by your ARTA displayed trace ( resized + reformatted ) for a more direct comparison to Bills work.
A few observations.
1) I would almost call these combos simply a 1 decade wide device when it comes to usable band-with response ( outside of the home environment ).
- meaning; a 1003b/288 combo will have an ( easy to use ) response of 300hz-3Khz ( before "Heroic" efforts kick in that are meant to achieve more extended HF response >> iow, drastic midrange attenuation ).
- likewise; the 805/288 combo looks like a 500-5Khz device ( before the "Heroics" ).
- I've never seen displayed; any meaningful trace of a 400 series multicell ( such as the 1004b ).
2) The HF "roll-off" characteristic ( for the 288/Altec multicell ) on the few traces I've seen seem to be pretty consistent at about 10db per octave.
- This steep a lowpass makes for real difficulty when designing a passive network for "flattish" response for the home environment ( a bit more later-on about this subject ).
3) The ( step-down ) low-frequency shelf ( below 1K in the 805b example ) is also is seen in the 1003b ( though starting a bit lower in frequency ) as well as is present in the large 329a radial ( though not as well defined to these eyes ).
- That lower octave "weakness" is also seen in the following trace of my own 288-8K on an Emilar EH500 pure radial ( though the LF is easily fixable using an underdamped HiPass filter to bump-up the bottom octave, before rolling off//out.
- This seems to be a common trait applying to all 288//horn combos ( either pure exponential multi-cells or a design mix > such as the radial ) > does your ES BiRadial also exhibit this trend ?? .
Here's the response after the HP ""bump" filter has been applied.
( this does sound very good >> even with all the "underdamped" areas present ( seen or unseen > such as a "Burst-Decay" response >> BTW; I consider "periods" more than 8 ( in the HF + UHF ) to be quite appropriate numbers as applied GaeQ ( for this fellow > GaeQ being simply "Geezer appropriate EQ"
).
- Note the "roughness" of the impedance curve below 3K ( I find this is typical when comparing any of the later "Tangerine" phase plugs to the more standard (W.E.) circumferential design types.
About that afore mentioned 10db per octave HF roll-off observed when standard Altec 288 drivers are mounted to Altec multi-cells.
It seems the "Hiraga" filter was based on a possible erroneous assumption that the HF roll-off ( lowpass ) was around 6db per octave.
One can see that this assumption likely started with the following excerpt lifted from John Stronczer's famed Sound Practices article ( from the 90's ).
The mentioned 9db drop over 1.5 octaves ( 5K to 15K ) equates to 6db per octave.
- The above pic certainly gives the impression that the 288K model has a 6db roll-off ( since that's the context that John S. is referring to, iow; his own gear ).
A more careful reading of that page shows that the outlined text refers to Hiraga's drivers, as measured in France ( iow; a Westrex 2080 driver fitted with an Altec 288 ohm diaphragm mounted to a 1505 multi-cell ).
- The paradox here, is I can't imagine that the Westrex driver ( patterned on the less efficient 288b+c models ) out-performs the later 288G model ( though it really seems to ). With the displayed trace being made with a CLIO system ( & even though it's an early model ) I have to believe the curves are/were well matched to the authors subjective impressions ( iow; I believe the displayed numbers are accurate ).
@Joseph Crowe ,
Here's the standard Hiraga HP filter applied to your 288-8g//805B;
The "red-line" is the Hiraga-response ( CLIO measurement ) superimposed ( as a target response ) for comparison to the predicted response of the 288g/805b combo ( displayed in green ).
- Black is your raw curve ( traced, less 5 db >> though I've added 2db back for this sim for added accuracy )
I find it really hard to reconcile the displayed FR curves to the stated filter used ( specifically, HP portion ) since the difference at 10K is a whopping 5db.
I know the Hiraga filter commonly under performs for many ( just ask Pano ) >> although there are others who are happy quite with it ).
A thoroughly tweaked Hiraga ( HP ) filter will bring the predicted response back to published expectations ( though I've had to add parts >> the greyed-out trace shows the effects of the newly added LCR notch-filter when it's "out-of-circuit" ).
See the following: ( the black > raw trace > had 2db added to what one sees here > to bring it into line with my very last trace );
Sharp eyes will see that this filter topology manages a couple db of UHF gain above 12.5K ( which is pretty useful ).
The 288-8K on the Altec MRii594a ( raw response as published by Altec for some reference for obtaining some of the levels I've used in my simms ).
Thanks for the data, all of it is much appreciated!
Question: Are you going to lash that 288-8G up to one of your ES BiRadials ( such as the ES 450 ) for a measurement session? I would love to see the results ( as I imagine would others ).
Also, I'm curious about what the impedance curve for the 288G/805b combo looks like ( could you post a pic of it please?? ).
Does it resemble the following ( iow; having a well defined peak of around 300hz )?? ( this pic lifted from Altec's own TL-231 technical letter (pdf) ).
Compare that to this earlier incarnation of the 288. This is a "C" model ( with a 16 ohm diaphragm ) in this case on a 329a Radial .
I do wonder if that impedance peak seen at 500hz is//was in fact accurate for that older diaphragm ( or possibly a mistake in the setup in the equipment for the combining of the twin traces on the B&K graph paper ).
Here's a Bill Woods ( RIP ) trace that he did of a 288C on a 1003b multicell.
- Followed by your ARTA displayed trace ( resized + reformatted ) for a more direct comparison to Bills work.
A few observations.
1) I would almost call these combos simply a 1 decade wide device when it comes to usable band-with response ( outside of the home environment ).
- meaning; a 1003b/288 combo will have an ( easy to use ) response of 300hz-3Khz ( before "Heroic" efforts kick in that are meant to achieve more extended HF response >> iow, drastic midrange attenuation ).
- likewise; the 805/288 combo looks like a 500-5Khz device ( before the "Heroics" ).
- I've never seen displayed; any meaningful trace of a 400 series multicell ( such as the 1004b ).
2) The HF "roll-off" characteristic ( for the 288/Altec multicell ) on the few traces I've seen seem to be pretty consistent at about 10db per octave.
- This steep a lowpass makes for real difficulty when designing a passive network for "flattish" response for the home environment ( a bit more later-on about this subject ).
3) The ( step-down ) low-frequency shelf ( below 1K in the 805b example ) is also is seen in the 1003b ( though starting a bit lower in frequency ) as well as is present in the large 329a radial ( though not as well defined to these eyes ).
- That lower octave "weakness" is also seen in the following trace of my own 288-8K on an Emilar EH500 pure radial ( though the LF is easily fixable using an underdamped HiPass filter to bump-up the bottom octave, before rolling off//out.
- This seems to be a common trait applying to all 288//horn combos ( either pure exponential multi-cells or a design mix > such as the radial ) > does your ES BiRadial also exhibit this trend ?? .
Here's the response after the HP ""bump" filter has been applied.
( this does sound very good >> even with all the "underdamped" areas present ( seen or unseen > such as a "Burst-Decay" response >> BTW; I consider "periods" more than 8 ( in the HF + UHF ) to be quite appropriate numbers as applied GaeQ ( for this fellow > GaeQ being simply "Geezer appropriate EQ"
).- Note the "roughness" of the impedance curve below 3K ( I find this is typical when comparing any of the later "Tangerine" phase plugs to the more standard (W.E.) circumferential design types.
About that afore mentioned 10db per octave HF roll-off observed when standard Altec 288 drivers are mounted to Altec multi-cells.
It seems the "Hiraga" filter was based on a possible erroneous assumption that the HF roll-off ( lowpass ) was around 6db per octave.
One can see that this assumption likely started with the following excerpt lifted from John Stronczer's famed Sound Practices article ( from the 90's ).
The mentioned 9db drop over 1.5 octaves ( 5K to 15K ) equates to 6db per octave.
- The above pic certainly gives the impression that the 288K model has a 6db roll-off ( since that's the context that John S. is referring to, iow; his own gear ).
A more careful reading of that page shows that the outlined text refers to Hiraga's drivers, as measured in France ( iow; a Westrex 2080 driver fitted with an Altec 288 ohm diaphragm mounted to a 1505 multi-cell ).
- The paradox here, is I can't imagine that the Westrex driver ( patterned on the less efficient 288b+c models ) out-performs the later 288G model ( though it really seems to ). With the displayed trace being made with a CLIO system ( & even though it's an early model ) I have to believe the curves are/were well matched to the authors subjective impressions ( iow; I believe the displayed numbers are accurate ).
@Joseph Crowe ,
Here's the standard Hiraga HP filter applied to your 288-8g//805B;
The "red-line" is the Hiraga-response ( CLIO measurement ) superimposed ( as a target response ) for comparison to the predicted response of the 288g/805b combo ( displayed in green ).
- Black is your raw curve ( traced, less 5 db >> though I've added 2db back for this sim for added accuracy )
I find it really hard to reconcile the displayed FR curves to the stated filter used ( specifically, HP portion ) since the difference at 10K is a whopping 5db.
I know the Hiraga filter commonly under performs for many ( just ask Pano ) >> although there are others who are happy quite with it ).
A thoroughly tweaked Hiraga ( HP ) filter will bring the predicted response back to published expectations ( though I've had to add parts >> the greyed-out trace shows the effects of the newly added LCR notch-filter when it's "out-of-circuit" ).
See the following: ( the black > raw trace > had 2db added to what one sees here > to bring it into line with my very last trace );
Sharp eyes will see that this filter topology manages a couple db of UHF gain above 12.5K ( which is pretty useful ).
The 288-8K on the Altec MRii594a ( raw response as published by Altec for some reference for obtaining some of the levels I've used in my simms ).
- The "K" version does poke above 110db ( by about 1 db ) which is why I added 2db ( for simming purposes ) to the 288-8g ( on a 805b multicell ) since it's apparently more efficient ( by 1db ) than the K versions.
- Altec shows the "K" version of the 288 having response over 10K ( that's about 102db )making that assertion good news for passive filter designers.
Last edited:
Also known as "waist banding", ~60 degree coverage at 1kHz.
With high frequency "fingering" 2.5kHz -12kHz.
That's a very interesting effect I was unaware of.
I had presumed the drop-offs were in the cell overlap zones, your directivity pattern chart clearly shows the shift from 1.5-3kHz "hot" in front of center cells, to the reverse above.
Nice testing report!
Did you perform a vertical directivity test?
Art