Hi all,
Nice work Vikash -
plenty enough information for me to get my thinking cap on.
As I don't have SW I couldn't do much with the files.
Boy the tweeter crossover is crude and a disaster.
As I suspected the bass unit Q is rather high,
giving the emphasis around 100 to 125 Hz.
Rb - there is an excellent article on negative impedance amplifiers
on the Elliot Sound Products website. Effectively what you are doing
is reducing the Re of the voicecoil and this decreases Qe value.
To do this you would need to rearrange the amplifier feedback loop.
However there are issues with driver stability at large excursion.
This route is possibly overkill for such a simple project, but I will
investigate changing the feeback loop on the main board, I can't
quite read the power amp IC identification but will find out.
Iwas going to sketch out the passive preamplifier but quickly realised
its basically a classic passive baxandall tone circuit + volume pot,
this has to go as its next to useless and loses a lot of signal.
I will be reworking this board into something I think is more useful.
Thanks all for all the help and comments so far.
/sreten.
Nice work Vikash -
plenty enough information for me to get my thinking cap on.
As I don't have SW I couldn't do much with the files.
Boy the tweeter crossover is crude and a disaster.
As I suspected the bass unit Q is rather high,
giving the emphasis around 100 to 125 Hz.
Rb - there is an excellent article on negative impedance amplifiers
on the Elliot Sound Products website. Effectively what you are doing
is reducing the Re of the voicecoil and this decreases Qe value.
To do this you would need to rearrange the amplifier feedback loop.
However there are issues with driver stability at large excursion.
This route is possibly overkill for such a simple project, but I will
investigate changing the feeback loop on the main board, I can't
quite read the power amp IC identification but will find out.
Iwas going to sketch out the passive preamplifier but quickly realised
its basically a classic passive baxandall tone circuit + volume pot,
this has to go as its next to useless and loses a lot of signal.
I will be reworking this board into something I think is more useful.
Thanks all for all the help and comments so far.
/sreten.
I have know about negative impedance for a while sreten was just curious to seeing those particular bits consort_ee_um was on about.
As negative impedance is out of the window, have you looked into QB5 alignments using a preceding active filter? These rely on a peaky box response so may be ideal in this situation.
If I was to have a go at this project I would be looking to add a dedicated amp for the tweeter. This would probably mean ripping the electronics out, and making a new plate amp with power active crossover, i.e. wrapping the filters round power amp chips such as TDA2050 or LM1875.
was just curious to seeing those particular bits consort_ee_um was on about.As negative impedance is out of the window, have you looked into QB5 alignments using a preceding active filter? These rely on a peaky box response so may be ideal in this situation.
If I was to have a go at this project I would be looking to add a dedicated amp for the tweeter. This would probably mean ripping the electronics out, and making a new plate amp with power active crossover, i.e. wrapping the filters round power amp chips such as TDA2050 or LM1875.
Hi,
my first thoughts are the passive preamp should include a low Q
notch filter centred on ~ 125Hz, this will improve the behaviour
in this region no end.
It could possibly also include a low Q high pass at say 50 to 60Hz.
The passive pre-amp should also include BSC, here I'm not sure
of the measurement conditions, in box or on a baffle / jig ?
Then its a case of smoothly integrating the upper mid and treble.
The passive pre-amp should also include some tone adjustment.
/sreten.
my first thoughts are the passive preamp should include a low Q
notch filter centred on ~ 125Hz, this will improve the behaviour
in this region no end.
It could possibly also include a low Q high pass at say 50 to 60Hz.
The passive pre-amp should also include BSC, here I'm not sure
of the measurement conditions, in box or on a baffle / jig ?
Then its a case of smoothly integrating the upper mid and treble.
The passive pre-amp should also include some tone adjustment.
/sreten.
Hi,
I've taken the measurements again today since I forgot to add the mic calibration data (posted graphs above have all been updated). I also needed to try it with an increased measurment repeat count. The net results aren't that different, but better to be complete...
I've taken the measurements again today since I forgot to add the mic calibration data (posted graphs above have all been updated). I also needed to try it with an increased measurment repeat count. The net results aren't that different, but better to be complete...
The first graph is of the woofer only. The 5 to 10k peak is the woofer.sreten said:
In box. Nearfield measurements used up to 300Hz and then spliced with farfield gated onaxis measurements (at approx 60cm) . All low end plots include combined port response.sreten said:
Hi,
if I've interpreted Vk's graphs correctly with cap c/o the tweeter is
effectively coming in at 10Khz - not too good. The range 5K to 10K
is all produced by the woofer, no wonder the upper mid doesn't
sound too hot.
The massive peak in the tweeters response around Fs in surprising,
around 12dB, suppressing this and getting reasonable output from
4/5k upwards is not going to be easy as the tweeter has no extra
efficiency advantage over the woofer we can utilise for filtering, it
has to be run full voltage above 5K.
On the overall response the dip just above 2K looks like phase
cancellation, so I'd say the target c/o point is 4Khz.
Ideal function for the tweeter is an elliptical filter - don't know
much about applying these to "hi-fi" units - but basically 2nd
order with a series notch filter.
This should get you to ~ 2nd order LR around 4Khz.
No point EQing the final octave of the the tweeter IMO.
So we need to add something like ~ 2nd order LR to the woofer,
probably need a 3rd order network to do this - and then hope
we've got reasonable phase matching or 180 degrees for the LR
summation (drivers at -6dB at c/o point).
/sreten.
if I've interpreted Vk's graphs correctly with cap c/o the tweeter is
effectively coming in at 10Khz - not too good. The range 5K to 10K
is all produced by the woofer, no wonder the upper mid doesn't
sound too hot.
The massive peak in the tweeters response around Fs in surprising,
around 12dB, suppressing this and getting reasonable output from
4/5k upwards is not going to be easy as the tweeter has no extra
efficiency advantage over the woofer we can utilise for filtering, it
has to be run full voltage above 5K.
On the overall response the dip just above 2K looks like phase
cancellation, so I'd say the target c/o point is 4Khz.
Ideal function for the tweeter is an elliptical filter - don't know
much about applying these to "hi-fi" units - but basically 2nd
order with a series notch filter.
This should get you to ~ 2nd order LR around 4Khz.
No point EQing the final octave of the the tweeter IMO.
So we need to add something like ~ 2nd order LR to the woofer,
probably need a 3rd order network to do this - and then hope
we've got reasonable phase matching or 180 degrees for the LR
summation (drivers at -6dB at c/o point).
/sreten.Good work Vikash and Sreten. I have just bought a pair and they are not bad. I reckon we will have them sounding like LS3/5A's before long.
RB Many moons ago I wrote an article for ETI - The Consort Loudspeaker which was a sixth order design with a negative output resistance amplifier. I went on to use Morel drive units in the design which had a superb effect.
A classic work on small active speakers was done by Recklinghausen, who came up with a sixth order design with 25dB of boost giving a 30Hz response from a 6 litre box! He backed off the boost when the woofer was going to hit the end stops, the so called dynamic EQ. This speaker was the KLH6, I wonder if any are still around. It would be fun to re-create the design, and see if the effects of dynamic EQ could be lived with.
RB Many moons ago I wrote an article for ETI - The Consort Loudspeaker which was a sixth order design with a negative output resistance amplifier. I went on to use Morel drive units in the design which had a superb effect.
A classic work on small active speakers was done by Recklinghausen, who came up with a sixth order design with 25dB of boost giving a 30Hz response from a 6 litre box! He backed off the boost when the woofer was going to hit the end stops, the so called dynamic EQ. This speaker was the KLH6, I wonder if any are still around. It would be fun to re-create the design, and see if the effects of dynamic EQ could be lived with.
They look about as old as you lot. [runs for cover]Hi all,
IMO 6th order negative impedance design is not appropriate
here, you need a better bass unit, and probably a lot more
amplifier power. I'd like to keep it as simple as possible with
the minimum of modification - an individual can always take
it further.
Current plans are to specify the rework to the preamp this weekend.
The bass knob will become a parametric EQ centred on 125Hz.
Range will be +/- 4dB with 0 the nominal upright position.
[Electrically 0dB(max) to - 8dB(min) with -4dB (centre)]
The treble knob will become a BSC circuit with 6dB step centre position.
Range will be no BSC (max position) to 12dB step (min position).
Volume knob as normal.
Still thinking about the c/o and possibly replacing the tweeter.
/sreten.
IMO 6th order negative impedance design is not appropriate
here, you need a better bass unit, and probably a lot more
amplifier power. I'd like to keep it as simple as possible with
the minimum of modification - an individual can always take
it further.
Current plans are to specify the rework to the preamp this weekend.
The bass knob will become a parametric EQ centred on 125Hz.
Range will be +/- 4dB with 0 the nominal upright position.
[Electrically 0dB(max) to - 8dB(min) with -4dB (centre)]
The treble knob will become a BSC circuit with 6dB step centre position.
Range will be no BSC (max position) to 12dB step (min position).
Volume knob as normal.
Still thinking about the c/o and possibly replacing the tweeter.
/sreten.Hi all,
I got sidetracked from the passive preamp with doing some
pink noise bands listening tests with the individual drivers.
Came to the conclusion I had 3 main problems whilst hopefully
the other purchasers will only have two.
1. The midbass unit midrange isn't the smoothest but eq'ing it
to any degree is not really practical. However there is a peak /
breakup mode around 6.3kHz - output above 8K is recessed
but remarkably smooth with no resonant problems.
2. On testing the tweeter I was coming to the conclusion it was
one of the worst sounding devices I've ever come across but on
checking the other tweeter output was much cleaner. In the
course of my investigations the front plate was removed and I
came to the conclusion the VC was rubbing the pole piece, at
which point I broke the VC wire from the terminal.
So off to RicherSounds for a pair of ten quid cheapo specials,
I'll have to persuade the similar tweeters into the original
faceplates, I'll report how similar the sound of the two are.
3. So back to the decent tweeter. Massive peak around 2K.
So the poor midrange I initially reported is the tweeter peak at
2K and the midbass peak at 6.3K and poor c/o integration.
My conclusions are :
The easiest c/o for the midbass unit is probably a notch filter
for the 6.3K peak and relying on the natural bass unit roll-off.
The tweeter needs a 2nd order c/o to suppress the 2K peak
and bring the response in flat rather than the first order tilted
response currently.
The actual acoustic c/o point will be around 7K. Not ideal in terms
of vertical resonse changes due to path differences, but IMO its
probably the best we can do.
Off to butcher the Ariston cheapo specials .........
/sreten.
I got sidetracked from the passive preamp with doing some
pink noise bands listening tests with the individual drivers.
Came to the conclusion I had 3 main problems whilst hopefully
the other purchasers will only have two.
1. The midbass unit midrange isn't the smoothest but eq'ing it
to any degree is not really practical. However there is a peak /
breakup mode around 6.3kHz - output above 8K is recessed
but remarkably smooth with no resonant problems.
2. On testing the tweeter I was coming to the conclusion it was
one of the worst sounding devices I've ever come across but on
checking the other tweeter output was much cleaner. In the
course of my investigations the front plate was removed and I
came to the conclusion the VC was rubbing the pole piece, at
which point I broke the VC wire from the terminal.
So off to RicherSounds for a pair of ten quid cheapo specials,
I'll have to persuade the similar tweeters into the original
faceplates, I'll report how similar the sound of the two are.
3. So back to the decent tweeter. Massive peak around 2K.
So the poor midrange I initially reported is the tweeter peak at
2K and the midbass peak at 6.3K and poor c/o integration.
My conclusions are :
The easiest c/o for the midbass unit is probably a notch filter
for the 6.3K peak and relying on the natural bass unit roll-off.
The tweeter needs a 2nd order c/o to suppress the 2K peak
and bring the response in flat rather than the first order tilted
response currently.
The actual acoustic c/o point will be around 7K. Not ideal in terms
of vertical resonse changes due to path differences, but IMO its
probably the best we can do.
Off to butcher the Ariston cheapo specials .........
/sreten.Sreten, having looked at the design I agree a negative resistance approach is over the top for this. However my measurements show that it is more of a boombox than Vikash measured. I was getting a 7.5dB peak with a nearfield measurement, on the non-amplified unit. I found that a bass tone setting control of 8.30 (clock setting, vertical being 12 o'clock) gave a flat response in the 100-300Hz region, although it is probably losing some bass at the low end. I got the flattest tone control/amplifier electrical response with both tone controls at 1 oclock. This is probably because the tone controls are asymetric that is having more cut than boost.
The amplifiers look like clones of the TDA2030A which is a SGS/Thomson design but it has someone else's logo
The tweeter capacitor is 2.2uF
The amplifiers look like clones of the TDA2030A which is a SGS/Thomson design but it has someone else's logo
The tweeter capacitor is 2.2uF
Hi all,
the Ariston cheapo specials were a result.
The tweeters have been persuaded into the original faceplates.
Audible response is very similar to the originals, if anything the
2.5K peak is a little higher, say 3K. They are 4 ohm units with
magnets ~ 50% bigger than the originals and output is higher,
I reckon with 3 ohm padding resistors they will be moreorless
indistinguishable from the originals.
The crossover and preamp details are delayed because
I'm having problems registering my circuit simulator.
/sreten.
the Ariston cheapo specials were a result.
The tweeters have been persuaded into the original faceplates.
Audible response is very similar to the originals, if anything the
2.5K peak is a little higher, say 3K. They are 4 ohm units with
magnets ~ 50% bigger than the originals and output is higher,
I reckon with 3 ohm padding resistors they will be moreorless
indistinguishable from the originals.
The crossover and preamp details are delayed because
I'm having problems registering my circuit simulator.
/sreten.Hi Sreten,
I tried plugging Vikash's values into Unibox and as you say a 4db peak results. Mind you I could persuade myself that Vikash is measuring 6db
So I tried varying values but no success in getting the 7.5db peak. I went for optimising the Fb to give a 7.5db peak which gave a value of 96.5Hz, but the tuning is clearly 72Hz. Perhaps some artefact of small boxes? To see what the negative resistance technique would give you try Qes=0.55.
I noticed the original tweeter is 8ohm which would cause level problems if we want to do any more than the original design. It seems a shame not to exploit the relatively flat response down to 4k
I tried plugging Vikash's values into Unibox and as you say a 4db peak results. Mind you I could persuade myself that Vikash is measuring 6db
So I tried varying values but no success in getting the 7.5db peak. I went for optimising the Fb to give a 7.5db peak which gave a value of 96.5Hz, but the tuning is clearly 72Hz. Perhaps some artefact of small boxes? To see what the negative resistance technique would give you try Qes=0.55.
I noticed the original tweeter is 8ohm which would cause level problems if we want to do any more than the original design. It seems a shame not to exploit the relatively flat response down to 4k
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.