the bass filter continued, the capacitors ...
continuing from my Post #655 of Page 66, with reference to Wayne Swann's Parts' list as on the layout diagram in #631 on Page 64.
The Bipolar capacitors in the two Parallel connected pairs of 33uF//39uF in the bass filter are Electrolytic capacitors,
and as such have significant Dielectric Absorption -{DA}.
Dielectric Absorption results in an amount of the signal's voltage being released over a period of time after the designed Time Constant of the filter,
and thus causes a smearing of the signal.
That is not High-Fidelity, and it will not allow the best that can be got from the Celestion bass drivers.
It makes no sense from a High-Fidelity stand-point to improve the inductors, and at significant cost, whilst installing new poor audio quality capacitors.
Celestion used bipolar electros in the original because for such large quantity of capacitance as required there were no cost-effective better large capacitors available then, and using multiple parallel connected smaller capacitors would have cost a very large amount of money, and occupied a very large amount of space, and thus the 66 would have been a very much more expensive l'speaker then.
The only quality benefit of those old bipolar electros was their sufficient ESR to part suppress the inevitable resonances caused by the high-slope filters, {both within themselves and with the drivers}.
Modern bipolar electros have less ESR, thus are of less use in that aspect,
and as their DA is not wanted -
except by listeners who like the sound effect caused by DA, and those listeners usually like the sound effect caused by high DCR inductors also
- there is no sense in installing new bipolar electros.
DA is inevitable in electrolytic caps - it cannot be reduced in the way that ESR can.
Polypropylene capacitors have very low DA, but are large in size, however they can be fitted to the existing board
- see moermusic's photos earlier in this thread -
and if one makes new boards to allow optimal positioning of the new low DCR inductors, such boards will be of sufficient size to fit the large polyproylene caps - in the spaces between the inductors.
It was recommended to Wayne Swann to use expensive Jantzen Silver Z caps in the treble filter.
The sound via those is completely wasted if bipolar electros are used in the bass filter.
A sensible proportioning of one's available budget is one which results in a similar sound quality across the entire bandwidth of the l'speaker.
Thus divide the budget proportional to the higher cost of the large components in the bass and low mids' filters, and then buy the best audio quality components that such will allow.
This will result in a balanced sound, and a balanced sound is audibly more satisfactory than even the best possible treble quality if the trade-off is smeared low mids and bass.
Remember, the low mids are being replayed via the bass drivers in the 66, thus affected by component quality there.
dloper reported the audible effects of this earlier in this thread,
and I have heard the same in various l'speakers.
If one wants to use Jantzen caps in the bass filter, the minimum quality level I recommend is to use their Cross Caps in pairs of 33uF//39uF.
Because the Cross Caps of 33uF and 39uF are the same length, they will have the same width of foil internally, thus will have the same Pulse Rise Time as each other, hence less distortion of the Transient Response of the signal than with parallel connected caps of different internal foil widths.
Because the two caps of the are not equal capacitance there will be a different Charge and Discharge time in each,
however that results in less audible problem than unequal Pulse Rise Time,
and for some listeners the sound caused as result of unequal capacitors is not an audible problem.
If you do not want unequal value capacitors, and I prefer not myself,
then use 36uF Solen or Axon{AEON} capacitors in pairs there.
Both brands are made in the same factory, though apparently slightly different construction for each, thus maybe slightly different sound between each.
Both are available in 250 volt size, thus physically smaller than the 400 volt size Jantzen Cross Caps, but if you want to use 400 volt rated Solen or Axon for the possible audible benefit which may result from the thicker dielectric of the 400 volt size, then that is not a problem if you have the space available.
A third option is to use single caps in each position, and for that I recommend you do what moermusic has done and use 75uF for the centre cap in the filter, and 68uF for the output cap.
See his Schematic in #659.
The larger portion of the filtering is done via the 3.5mH inductor, thus the lower impedance of the 75uF cap is of use there,
and the 68uF cap is sufficient to follow the 2.2mH filter inductor.
This will average out at almost identical measured response to that if two 72uF caps were used, and is most likely audibly indistinguishable from such.
I will discuss the midrange filter next time,
but do post and query anything about the bass filter now if such is not sufficiently explained here-in.
continuing from my Post #655 of Page 66, with reference to Wayne Swann's Parts' list as on the layout diagram in #631 on Page 64.
The Bipolar capacitors in the two Parallel connected pairs of 33uF//39uF in the bass filter are Electrolytic capacitors,
and as such have significant Dielectric Absorption -{DA}.
Dielectric Absorption results in an amount of the signal's voltage being released over a period of time after the designed Time Constant of the filter,
and thus causes a smearing of the signal.
That is not High-Fidelity, and it will not allow the best that can be got from the Celestion bass drivers.
It makes no sense from a High-Fidelity stand-point to improve the inductors, and at significant cost, whilst installing new poor audio quality capacitors.
Celestion used bipolar electros in the original because for such large quantity of capacitance as required there were no cost-effective better large capacitors available then, and using multiple parallel connected smaller capacitors would have cost a very large amount of money, and occupied a very large amount of space, and thus the 66 would have been a very much more expensive l'speaker then.
The only quality benefit of those old bipolar electros was their sufficient ESR to part suppress the inevitable resonances caused by the high-slope filters, {both within themselves and with the drivers}.
Modern bipolar electros have less ESR, thus are of less use in that aspect,
and as their DA is not wanted -
except by listeners who like the sound effect caused by DA, and those listeners usually like the sound effect caused by high DCR inductors also
- there is no sense in installing new bipolar electros.
DA is inevitable in electrolytic caps - it cannot be reduced in the way that ESR can.
Polypropylene capacitors have very low DA, but are large in size, however they can be fitted to the existing board
- see moermusic's photos earlier in this thread -
and if one makes new boards to allow optimal positioning of the new low DCR inductors, such boards will be of sufficient size to fit the large polyproylene caps - in the spaces between the inductors.
It was recommended to Wayne Swann to use expensive Jantzen Silver Z caps in the treble filter.
The sound via those is completely wasted if bipolar electros are used in the bass filter.
A sensible proportioning of one's available budget is one which results in a similar sound quality across the entire bandwidth of the l'speaker.
Thus divide the budget proportional to the higher cost of the large components in the bass and low mids' filters, and then buy the best audio quality components that such will allow.
This will result in a balanced sound, and a balanced sound is audibly more satisfactory than even the best possible treble quality if the trade-off is smeared low mids and bass.
Remember, the low mids are being replayed via the bass drivers in the 66, thus affected by component quality there.
dloper reported the audible effects of this earlier in this thread,
and I have heard the same in various l'speakers.
If one wants to use Jantzen caps in the bass filter, the minimum quality level I recommend is to use their Cross Caps in pairs of 33uF//39uF.
Because the Cross Caps of 33uF and 39uF are the same length, they will have the same width of foil internally, thus will have the same Pulse Rise Time as each other, hence less distortion of the Transient Response of the signal than with parallel connected caps of different internal foil widths.
Because the two caps of the are not equal capacitance there will be a different Charge and Discharge time in each,
however that results in less audible problem than unequal Pulse Rise Time,
and for some listeners the sound caused as result of unequal capacitors is not an audible problem.
If you do not want unequal value capacitors, and I prefer not myself,
then use 36uF Solen or Axon{AEON} capacitors in pairs there.
Both brands are made in the same factory, though apparently slightly different construction for each, thus maybe slightly different sound between each.
Both are available in 250 volt size, thus physically smaller than the 400 volt size Jantzen Cross Caps, but if you want to use 400 volt rated Solen or Axon for the possible audible benefit which may result from the thicker dielectric of the 400 volt size, then that is not a problem if you have the space available.
A third option is to use single caps in each position, and for that I recommend you do what moermusic has done and use 75uF for the centre cap in the filter, and 68uF for the output cap.
See his Schematic in #659.
The larger portion of the filtering is done via the 3.5mH inductor, thus the lower impedance of the 75uF cap is of use there,
and the 68uF cap is sufficient to follow the 2.2mH filter inductor.
This will average out at almost identical measured response to that if two 72uF caps were used, and is most likely audibly indistinguishable from such.
I will discuss the midrange filter next time,
but do post and query anything about the bass filter now if such is not sufficiently explained here-in.
Last edited:
correction to the above post
I am surprised that no-one has challenged me about this, because as it reads it is incorrect in this particular circuit.
I didn't have time to Proof-Read last time, thus didn't see that and correct it.
Capacitors connected in Parallel will all charge at the same rate, because the current through each whilst charging will be proportional to the Impedance -{the geometric sum of capacitive reactance and ESR and ESL} of each.
The Pulse Rise Time through each will be determined by the width of the internal foil, thus why a shorter capacitor passes high frequencies faster than a longer capacitor, but both would fully charge in the same time if their ESRs and ESLs are equivalently proportional to their capacitive reactances,
{and that is usually not the case for two caps of significantly different internal foil widths, and particually not if one cap is significantly greater capacitance than the other because it will likely have unproportionally greater ESL}.
Some listeners have reported hearing a difference between two equal value caps connected in Parallel compared to two unequal values of the same length,
eg: as would be the case comparing the equal length 33uF//39uF with 36uF//36uF,
and that has been claimed even for two nominally equal value caps when one has been found to be significantly + in its Tolerance and the other significantly -.
{Tolerance is the capacitance variation from the nominal value, and is usually +/- 10% for modern plastic film caps,
and quite often +/- 5% for the specialist polypropylene types made for cross-overs.}
Thus with +/- 5% caps any two might be different by as much as 37.8uF and 34.2uF for the 36uF pair.
I have not experimented with this with large value caps used in the Parallel arms of bass filters, however I doubt a same brand/model, equal lengths' 36//36 if actual capacitances were 34.2 and 37.8 would cause audible problems here.
I would only expect this effective 10% difference to cause an audible annomaly with upper midrange and treble frequencies,
thus if one has to use a several parallel connected caps to sum to 24uF for the mids' filter cap of the 66's crossover, then pay the seller to measure and select close to pairs of 12uF or threes of 8.2uF, or buy a 25uF Clarity Cap, or a Sonicap.
As for 33uF//39uF, well if equal length, same brand and type, then this is not likely to be a problem in the bass filter,
but if you are worried then use 75uF for the centre cap and 68uF for the output caps, as moermusic has done.
If you already have caps, and they are different lengths, then post their values here.
They can be made to perform similarly by using a separate ESR resistor of different resistance with each cap instead of a single resistor connected to one end of the pair.
Yes, the ESR simulating resistors are still necessary in the bass filter.
These do NOT attenuate useful signal to the woofer.
They are Damping resistors to suppress the inevitable resonances caused by the new low ESR caps interacting with the inductance of the woofer's voice-coil,
in addition to suppressing the internal resonance of the filter itself as now less internal resistance in the filter if one installs new low DCR inductors.
When one attempts to improve one aspect of a filter, one has to take care to not introduce a new problem in the filter.
I am surprised that no-one has challenged me about this, because as it reads it is incorrect in this particular circuit.
I didn't have time to Proof-Read last time, thus didn't see that and correct it.
Capacitors connected in Parallel will all charge at the same rate, because the current through each whilst charging will be proportional to the Impedance -{the geometric sum of capacitive reactance and ESR and ESL} of each.
The Pulse Rise Time through each will be determined by the width of the internal foil, thus why a shorter capacitor passes high frequencies faster than a longer capacitor, but both would fully charge in the same time if their ESRs and ESLs are equivalently proportional to their capacitive reactances,
{and that is usually not the case for two caps of significantly different internal foil widths, and particually not if one cap is significantly greater capacitance than the other because it will likely have unproportionally greater ESL}.
Some listeners have reported hearing a difference between two equal value caps connected in Parallel compared to two unequal values of the same length,
eg: as would be the case comparing the equal length 33uF//39uF with 36uF//36uF,
and that has been claimed even for two nominally equal value caps when one has been found to be significantly + in its Tolerance and the other significantly -.
{Tolerance is the capacitance variation from the nominal value, and is usually +/- 10% for modern plastic film caps,
and quite often +/- 5% for the specialist polypropylene types made for cross-overs.}
Thus with +/- 5% caps any two might be different by as much as 37.8uF and 34.2uF for the 36uF pair.
I have not experimented with this with large value caps used in the Parallel arms of bass filters, however I doubt a same brand/model, equal lengths' 36//36 if actual capacitances were 34.2 and 37.8 would cause audible problems here.
I would only expect this effective 10% difference to cause an audible annomaly with upper midrange and treble frequencies,
thus if one has to use a several parallel connected caps to sum to 24uF for the mids' filter cap of the 66's crossover, then pay the seller to measure and select close to pairs of 12uF or threes of 8.2uF, or buy a 25uF Clarity Cap, or a Sonicap.
As for 33uF//39uF, well if equal length, same brand and type, then this is not likely to be a problem in the bass filter,
but if you are worried then use 75uF for the centre cap and 68uF for the output caps, as moermusic has done.
If you already have caps, and they are different lengths, then post their values here.
They can be made to perform similarly by using a separate ESR resistor of different resistance with each cap instead of a single resistor connected to one end of the pair.
Yes, the ESR simulating resistors are still necessary in the bass filter.
These do NOT attenuate useful signal to the woofer.
They are Damping resistors to suppress the inevitable resonances caused by the new low ESR caps interacting with the inductance of the woofer's voice-coil,
in addition to suppressing the internal resonance of the filter itself as now less internal resistance in the filter if one installs new low DCR inductors.
When one attempts to improve one aspect of a filter, one has to take care to not introduce a new problem in the filter.
Last edited:
DennyG's case with all 68uF caps
Hi DennyG,
I think PB series Solen are the 400 volt rated, thus will be large physical size.
With 68uF in both positions of the bass filter there may not be sufficiently low parallel impedance for the filter to cut off as low in frequency as needed.
Given the Solens could be anywhere between 64.6uF and 71.4uF,
when you disconnect them to add the series resistors measure the capacitance of each.
If you don't have a meter to do this with, then find some-one who has a Multimeter with a Capacitance range or an LCR {Inductance/Capacitance/Resistance} meter,
and pay them to measure for you.
Label each cap with its measured result so that you know which is which,
and post the measured values here and I will advise if any changes should be made.
Perhaps a small value cap in parallel with the centre position 68uF cap, and with its own separate resistor, will suffice.
My cousin in Australia informed me that Jaycar there sell:
Inductance/Capacitance/Frequency meter QM-1324, which also measures Resistance, for less than AUD50,
but if you don't need a meter for other things and cost of measurement seems excessive, then simpler would be to buy two 75uF caps
- one for each crossover.
If you have to buy those from overseas, then maybe when you decide to upgrade the other caps, if ...
What do your 66s sound like now with the resistors installed ?
Hi DennyG,
I think PB series Solen are the 400 volt rated, thus will be large physical size.
With 68uF in both positions of the bass filter there may not be sufficiently low parallel impedance for the filter to cut off as low in frequency as needed.
Given the Solens could be anywhere between 64.6uF and 71.4uF,
when you disconnect them to add the series resistors measure the capacitance of each.
If you don't have a meter to do this with, then find some-one who has a Multimeter with a Capacitance range or an LCR {Inductance/Capacitance/Resistance} meter,
and pay them to measure for you.
Label each cap with its measured result so that you know which is which,
and post the measured values here and I will advise if any changes should be made.
Perhaps a small value cap in parallel with the centre position 68uF cap, and with its own separate resistor, will suffice.
My cousin in Australia informed me that Jaycar there sell:
Inductance/Capacitance/Frequency meter QM-1324, which also measures Resistance, for less than AUD50,
but if you don't need a meter for other things and cost of measurement seems excessive, then simpler would be to buy two 75uF caps
- one for each crossover.
If you have to buy those from overseas, then maybe when you decide to upgrade the other caps, if ...
What do your 66s sound like now with the resistors installed ?
Last edited:
Capacitance measuring
The inductance capability should be useful as well. My current meter has a limited capacitance range and no inductance capability.
---------------
Hi DennyG,
as you stated the above, I reply:
as you have a Capacitance measuring range it can most likely be used along with an additional capacitor to effectively measure the 68uF caps.
One of the other caps in the crossover may suffice for that purpose.
I can describe how to do that, though to do it succinctly {me, succinct !},
post here the Maximum value of capacitance your meter is specified to measure ?
I see the Jaycar meter is specified for +/- 2.5% accuracy for Capacitance and Inductance, which is OK for making relative measurements, but not accurate enough to be sure if your 68uF caps are + or - of their nominal value.
Similar would be the case with unknown inductors.
If you need an LCR meter, it would be better to buy one with a +/- 1%, at least, accuracy for the capacitance and inductance,
{though that will likely be double the price}.
The basic +/- 1% spec of the Jaycar meter is only for its Resistance and Frequency ranges, which is not bad,
but those will be +/- 0.5% at least with a better meter, and that is more useful ... particually if the meter has only limited Resolution ... eg: if 3 + 1/2 Digits Scale.
Look in the booklet for your current meter, and you will see different accuracies for different functions, and for some functions the accuracy may be different for different measurement ranges ... that is inevitable with Digital meters.
What is the Accuracy spec for the largest capacitance range of your meter ?
***************
For their current ranges, Solen state PA series is 250 VDC and PB series is 400 VDC.
Either will work, so don't worry too much now,
but wait till you hear what the sound is like after sufficient period for your ear/brain to have adjusted for the change.
By that I mean don't try to hear differences, because you will notice over a period of time if any real differences are there.
The inductance capability should be useful as well. My current meter has a limited capacitance range and no inductance capability.
---------------
Hi DennyG,
as you stated the above, I reply:
as you have a Capacitance measuring range it can most likely be used along with an additional capacitor to effectively measure the 68uF caps.
One of the other caps in the crossover may suffice for that purpose.
I can describe how to do that, though to do it succinctly {me, succinct !},
post here the Maximum value of capacitance your meter is specified to measure ?
I see the Jaycar meter is specified for +/- 2.5% accuracy for Capacitance and Inductance, which is OK for making relative measurements, but not accurate enough to be sure if your 68uF caps are + or - of their nominal value.
Similar would be the case with unknown inductors.
If you need an LCR meter, it would be better to buy one with a +/- 1%, at least, accuracy for the capacitance and inductance,
{though that will likely be double the price}.
The basic +/- 1% spec of the Jaycar meter is only for its Resistance and Frequency ranges, which is not bad,
but those will be +/- 0.5% at least with a better meter, and that is more useful ... particually if the meter has only limited Resolution ... eg: if 3 + 1/2 Digits Scale.
Look in the booklet for your current meter, and you will see different accuracies for different functions, and for some functions the accuracy may be different for different measurement ranges ... that is inevitable with Digital meters.
What is the Accuracy spec for the largest capacitance range of your meter ?
***************
For their current ranges, Solen state PA series is 250 VDC and PB series is 400 VDC.
Either will work, so don't worry too much now,
but wait till you hear what the sound is like after sufficient period for your ear/brain to have adjusted for the change.
By that I mean don't try to hear differences, because you will notice over a period of time if any real differences are there.
Last edited:
Hi Alan,
What improvement in sound quality do you believe this change would make? The discussions I have had with others do not agree that this would make considerable improvement to the sound quality.
When I approached this upgrade project I went ahead planing improvements to the crossovers placing them under the following categories:-
Budget - The lowest price to make a significant improvement in sound.
Mid-range - A good balance. Where spending more brings less obvious benefit in sound quality.
High-end - Money no object.
In the end I decided to go with the changes in the 'Mid-Range' category along with a couple of components from the 'High-End' category. I understand that other improvements could be made but they come at too greater cost for the percentage/amount of improvement in sound quality. I have to say that the changes I have made using the previous layout I have shown and also relocating the crossover from behind the large magnet of the bass driver have really brought the sound of these speakers into the 21sy Century! they sound fantastic!?
How would you list changes to the crossovers using the same categories of Cost vs Improvement/Benefit?
Best Regards
Wayne
Last edited:
Hi again Alan,
I would like to add that the upgrade I have made so far will do me for a while, but I do expect to improve it eventually when money isn't so tight. I would therefore still be very grateful for the comments you have regarding the ideal improvements that could be made. I must also add that after reading my previous entry it may have seemed rather disrespectful which was not my intent at all. I tried changing it but ran out of the 30 minute time limit.
Best Regards
Wayne
I would like to add that the upgrade I have made so far will do me for a while, but I do expect to improve it eventually when money isn't so tight. I would therefore still be very grateful for the comments you have regarding the ideal improvements that could be made. I must also add that after reading my previous entry it may have seemed rather disrespectful which was not my intent at all. I tried changing it but ran out of the 30 minute time limit.
Best Regards
Wayne
Last edited:
Hi,
If you look back to the top of page 64 of this Thread, entry #661, you will see what the crossover looks like and the associated schematic.
Best Regards
Wayne
Last edited:
20uF and +-4%.
I'll look around for one with better accuracy Alan.
Jaycar have one on their web site (QM1535) which they say has:
• Capacitance: 4nF, 40nF, 400nF, 4µ, 40µF
• Base accuracy: ±0.5% of reading
but at $35 this is hard to believe.
Jaycar also have a dedicated capacitance meter which may be a more reliable option accuracy wise:
Digital Capacitance Meter
- 0.1PF TO 20,000µF
- HOLSTER INCLUDED
The Digitech QM-1572 includes a yellow holster. Features include 21mm LCD display, 9 ranges from 200pf to 20mf, high accuracy and zero adjust display.
- Includes leads - banana plugs to alligator clips.
Specifications:
Normal Max in range
Resolution Test
Range Display
200pF 199.9pF 0.1pF 800Hz
2nF 1.999nF 1pF 800Hz
20nF 19.99nF 10pF 800Hz
200nF 199.9nF 100pF 800Hz
2uF 1.999uF 1000pF 800Hz
20uF 19.99uF 0.01uF 80Hz
200uF 199.9uF 0.1uF 8Hz
2000uF 1999uF 1uF 8Hz
20,000uF 19,999uF,10uF 8Hz
- Accuracy+/-0.5% on 200pf to 200uf ranges+/-1% on 2,000uf range +/-2% on 20mf range
- Excitation voltage 2.8 volts peak
- Battery 9V
- Size 191(L) x 89(W) x 35(H)mm
QTY 1+ $59.95 3+ $53.95 6+ $47.90
to measure close to the capacitance, + interim to Wayne Swann
Hi DennyG,
with your meter to 20uF it is possible to measure larger capacitors.
To do this you will need a plastic film cap, {such as a Polypropylene is}, of no larger than 27uF, and preferably of +/- 5%, or better, Tolerance.
Thus if there is a Solen 27uF, +/- 5% in your crossover, that can be used.
If you don't have 27uF, then use the LARGEST capacitor you have that is less than 27uF.
There may not be a +/- % printed on some poly caps, but instead will be printed a letter, such as K or J or G, etc ... after the capacitance code digits.
The closer to the beginning of Alphabet this letter is, the better the % tolerance of the capacitor.
Scrape clean part of the leads of each cap to be measured, including the additional cap.
I will refer here-after to the additional cap as the Calc-cap.
Connect the Calc-cap in electrical Series with one 68uF cap,
simply with an Alligator clip holding the two leads together at the part of each that you scraped clean.
Connect your meter's leads across the complete pair, as below:
Lead --> Calc-cap + 68uF cap <-- Lead
With a 27uF Calc-cap, if the 68uF cap happens to be 70uF, the meter will read 20uF.
If the meter indicates an "over scale", then either the Test cap is greater than 70uF or the Calc-cap is greater than 27uF.
If you are using smaller than 27uF for Calc-cap, there will not be any "over scale" unless one of the caps is faulty.
The accuracy is better with larger rather than smaller for the Calc-cap,
though with your meter it can never be better than +/- 4%, however that will give us some idea of where these 68uF caps are.
These large caps will draw a lot of current from the meter's battery,
thus install a new battery if yours has been used for a while or the reading may be faulty.
Measure quickly, removing the test leads as soon as the digits' reading on the meter have stabilized, so that you are not depleting the battery.
Label each 68uF cap so that you know which is which.
Post here the 4 measured results AND the nominal capacitance of your Calc-cap, and its Tolerance letter, or its +/- %.
If you want to calculate the results yourself, the formula is:-
uF of Calc-cap multiplied meas'd uF, divided by meas'd uF minus uF of Calc-cap,
that is:- C1 x C2/{C2 - C1} = C of the 68uF cap.
***************
Note that for the range above 20uF the test signal is 8 Hz.
That is significantly below the frequencies we want to know the capacitance,
thus accuracy will not be quite as good as +/- 0.5%,
however it is not bad, and is still better than the case will be with your meter.
For Capacitance and Inductance measurements it is better if the test frequencies are in the general area of the frequency bands the components will be used in,
thus for that meter the ranges with 80Hz and 800 Hz test signals are more useful.
I would only buy that meter if I wanted all those cap ranges for not expensive price, but we may get useable results with the method I described using your meter, thus try that first.
Buying any meter, one should consider what else one may want to measure,
and for crossovers it is useful to have Inductance as well,
but if you are never going to measure again after this project it would be better to spend the money on the best capacitors, and maybe inductors, than on a meter.
Of course, if you buy a suitable meter you measure Inductors that you wind for yourself ... if you want to take the time to wind inductors !
***************
Hi Wayne Swann,
I will return when I have time available to comment further about your specific circuit,
but for now I recommend you listen with as wide a variety of types of recordings that you own - as many different types of Instruments and Voices as possible, and in Acoustic and Electric styles, and Classical Music -{unless you only prefer one type}.
Listen to some older recordings also, not only ones from the recent 20 years.
Hi DennyG,
with your meter to 20uF it is possible to measure larger capacitors.
To do this you will need a plastic film cap, {such as a Polypropylene is}, of no larger than 27uF, and preferably of +/- 5%, or better, Tolerance.
Thus if there is a Solen 27uF, +/- 5% in your crossover, that can be used.
If you don't have 27uF, then use the LARGEST capacitor you have that is less than 27uF.
There may not be a +/- % printed on some poly caps, but instead will be printed a letter, such as K or J or G, etc ... after the capacitance code digits.
The closer to the beginning of Alphabet this letter is, the better the % tolerance of the capacitor.
Scrape clean part of the leads of each cap to be measured, including the additional cap.
I will refer here-after to the additional cap as the Calc-cap.
Connect the Calc-cap in electrical Series with one 68uF cap,
simply with an Alligator clip holding the two leads together at the part of each that you scraped clean.
Connect your meter's leads across the complete pair, as below:
Lead --> Calc-cap + 68uF cap <-- Lead
With a 27uF Calc-cap, if the 68uF cap happens to be 70uF, the meter will read 20uF.
If the meter indicates an "over scale", then either the Test cap is greater than 70uF or the Calc-cap is greater than 27uF.
If you are using smaller than 27uF for Calc-cap, there will not be any "over scale" unless one of the caps is faulty.
The accuracy is better with larger rather than smaller for the Calc-cap,
though with your meter it can never be better than +/- 4%, however that will give us some idea of where these 68uF caps are.
These large caps will draw a lot of current from the meter's battery,
thus install a new battery if yours has been used for a while or the reading may be faulty.
Measure quickly, removing the test leads as soon as the digits' reading on the meter have stabilized, so that you are not depleting the battery.
Label each 68uF cap so that you know which is which.
Post here the 4 measured results AND the nominal capacitance of your Calc-cap, and its Tolerance letter, or its +/- %.
If you want to calculate the results yourself, the formula is:-
uF of Calc-cap multiplied meas'd uF, divided by meas'd uF minus uF of Calc-cap,
that is:- C1 x C2/{C2 - C1} = C of the 68uF cap.
***************
Note that for the range above 20uF the test signal is 8 Hz.
That is significantly below the frequencies we want to know the capacitance,
thus accuracy will not be quite as good as +/- 0.5%,
however it is not bad, and is still better than the case will be with your meter.
For Capacitance and Inductance measurements it is better if the test frequencies are in the general area of the frequency bands the components will be used in,
thus for that meter the ranges with 80Hz and 800 Hz test signals are more useful.
I would only buy that meter if I wanted all those cap ranges for not expensive price, but we may get useable results with the method I described using your meter, thus try that first.
Buying any meter, one should consider what else one may want to measure,
and for crossovers it is useful to have Inductance as well,
but if you are never going to measure again after this project it would be better to spend the money on the best capacitors, and maybe inductors, than on a meter.
Of course, if you buy a suitable meter you measure Inductors that you wind for yourself ... if you want to take the time to wind inductors !
***************
Hi Wayne Swann,
I will return when I have time available to comment further about your specific circuit,
but for now I recommend you listen with as wide a variety of types of recordings that you own - as many different types of Instruments and Voices as possible, and in Acoustic and Electric styles, and Classical Music -{unless you only prefer one type}.
Listen to some older recordings also, not only ones from the recent 20 years.
Hi Alan,
Thanks for your comments. I will certainley be listening to as much varied music I can. I do have of course several favourites which I have listened to over the years as my Hi-Fi system has grown in sophistication. Hopefully it'll give me a good idea of what the new Celestion 66's are capable of in their new
(I just about finished this reply and lost it all somehow so I'll cut to the chase)
Thanks Alan.
My old meter proved inconvenient to use in this situation as the caps need to be inserted in closely spaced narrow slots - can't use the test leads. This, the range issue and the poor accuracy led me to get the C meter. Now I can be confident in the C measurements. You're right, I don't intend to wind inductors!
Note that the Solens in my 66 crossovers have the following nominal values:
3.9uF (4uF in original)
6.2uF (6uF in original)
30uF (same as original)
68uF (72uF in original)
All the Solens I have measured read slightly under the nominal value.
eg the 68uFs read: 66.3uF, 67.5uF, 67.6uF, 67.6uF
I welcome your opinion on these values as only the 30uF is identical to the caps in the original design.
I also note that Celestion reduced to 30uF to 24uF in the later models. If I get new caps I think I'll go for 24uF. I remember a friend had a pair of 66's that were purchased a few years later than mine and they probably sounded better than the earlier model but we never did an A/B on them with the same equipment in the same room to verify that. The speakers were too awkward and heavy to lug around. We just enjoyed the music.
next step
Hi DennyG,
I think with both bass filter caps at about 4% less than Celestion specification there will be not quite enough for optimum filtering, but do not worry because later you can effectively increase one, or both, in each crossover by using you Solen 6u2 and 3u9 caps in Parallel with the 68(-)s if you decide to upgrade to another brand for mids and treble.
First step is to listen with the new resistors in circuit
- in the positions as moermusic has drawn for us in #659 on Page 66 -
noting that you have 2.7 ohm and not 3.3 ohm for one of them.
If you have bought the 82 ohm/10 watt wirewounds, then include those also, but if you have not yet bought those, then buy 68 ohm instead, because you have 30uF caps, thus to relieve your old mid-domes of excess low midrange excursion it will be slightly better if the 30uF caps see a slightly lower Impedance.
Do not buy expensive 68 ohm - buy cheap 10 watt wirewounds from the local parts shop, because this is for interim use only until you hear what you want to do next.
If you decide to upgrade the caps, then it will be better to buy 24uF or 25uF for the mids, and then you can buy good quality 82 ohm resistors to use with those.
If you have the 68uF caps out of circuit, then put the two largest measured in the output positions - in Parallel with the woofers -
and the two lower meas'd in the centre position of the T shape formed with the Inductors.
I have to go now.
Hi DennyG,
I think with both bass filter caps at about 4% less than Celestion specification there will be not quite enough for optimum filtering, but do not worry because later you can effectively increase one, or both, in each crossover by using you Solen 6u2 and 3u9 caps in Parallel with the 68(-)s if you decide to upgrade to another brand for mids and treble.
First step is to listen with the new resistors in circuit
- in the positions as moermusic has drawn for us in #659 on Page 66 -
noting that you have 2.7 ohm and not 3.3 ohm for one of them.
If you have bought the 82 ohm/10 watt wirewounds, then include those also, but if you have not yet bought those, then buy 68 ohm instead, because you have 30uF caps, thus to relieve your old mid-domes of excess low midrange excursion it will be slightly better if the 30uF caps see a slightly lower Impedance.
Do not buy expensive 68 ohm - buy cheap 10 watt wirewounds from the local parts shop, because this is for interim use only until you hear what you want to do next.
If you decide to upgrade the caps, then it will be better to buy 24uF or 25uF for the mids, and then you can buy good quality 82 ohm resistors to use with those.
If you have the 68uF caps out of circuit, then put the two largest measured in the output positions - in Parallel with the woofers -
and the two lower meas'd in the centre position of the T shape formed with the Inductors.
I have to go now.
(1a) - this will NOT give a clean modern version of the Ditton 66 sound,
because of the significant changes to the circuit, especially those to the low-pass filter section of the midrange circuit.
(1b) - why have you proposed that change to that part of the filter ?
(2) - it will sound excessive and forward in the midrange, because there are no attenuating resistors in what is a very Resonant circuit there.
(3) - the mid-dome may be damaged, or at least have its life reduced faster over time when the speakers are played loud, because of the 27uF capacitor.
Despite there being 30uF in the first version of the Ditton 66, Celestion reduced that to 24uF for good reason - to not permit excess low mids to drive the mid-dome.
(4a) - it will sound compressed and weak in the bass when driven hard, because of those cored inductors you have indicated.
Why are you changing from the original air-core inductors ?
(4b) - are those small-drawn caps in the bass filter plastic film or electrolytic ?
(5) - tweeter response is hard to predict straight away, but I'll study the circuit and reply further when I've evaluated the treble filter.
(6) - are those horizontally placed L1/.11mH and L3/.8mH inductors also solid-core of some type ? ...
and if so, why ?
(7) - it is a waste of money to buy the expensive Jantzen "Silver" caps if they are put into a circuit which is not optimised for balanced coherent sound.
Expensive capacitors cannot improve a flawed circuit design.
You can of course buy the components, and build this crossover - that is your right, and perhaps you will like it, but many have gone down similar paths and regretted it.
It looks very unbalanced to me, thus I'm interested to hear your reasons for proposing it ..?
__________________
Alan
Originally Posted by alan-1-b
Hi Wayne,
It looks very unbalanced to me, thus I'm interested to hear your reasons for proposing it ..?
Attached FilesAn externally hosted image should be here but it was not working when we last tested it.New crossover configuration.doc (180.0 KB, 21 views)
Ask these things: 
Hi Wayne,
very interesting, but the provided explanations omit some relevant matters,
especially about the Bass performance !
however, before we get to that, I recommend you ask your circuit designer the following:
(1) - why is the C9 - L8 - R1 network in Parallel with the SEAS tweeter tuned to 5571 Hz ?
I would have thought it should be tuned to the tweeter's Fs, which is aabout 1600 Hz ... for which C9 would be about 8.2uF.
Perhaps there is an error there ..?
(2) - regardless whether the C9/L8/R1 network is correct or not, L1 is a very low value, and in combination with C3 and C1 will cause a lot of loss around the 5kHz region.
Why has that been done ?
(3) - despite the mention of Impedance Correction for the MD500, there is not any impedance correction circuit included for it, and especially not at the low frequency end where such could be most beneficial.
Why is it not included ?
(4) - the annomolies in the upper midrange Frequency Response of the MD500s - are these YOUR samples of MD500s, or other samples ?
As sba showed with his measurements, these old mid-domes do have annomolies, AND those vary between samples, thus whose Samples are shown in your measurements ?
(5) - in the midrange filter, L3 and C5 will cause substantial loss of output in the region just below 5kHz, thus when combined with the seeming large loss in the same area of the frequency spectrum caused by L1/C3/C5 in the treble filter, I think it likely there will be a hole in the frequency response there - a very audible one.
Perhaps there is an error in the calculations, or in the printed results - ask that to be checked ..?
(6) - the Resistors I recommended are NOT there to degrade the performance of the new capacitors {to make them sound like old capacitors},
they are there to keep the midrange in coherent balance with the treble and bass that is audible in this loudspeaker.
The 66 is a Balance of Compromises, and some of that I explained in various places during this Thread.
All loudspeakers require a Balance of Compromises, and if you want to update the sound using the old ABRs and woofers and mid-domes you will have to use another type of Balance of Compromises , otherwise some parts of the frequency spectrum will not be coherently in time with others when music is played ,{as distinct from steady state test tones}, - in other words, it can sound like a wobbly mess.
Enquire Points (1) <---> (5) inclusive first, and Post the replies, and then I will discuss the Bass issues ... if there seems to be any sense in proceeding with this design.
__________________
Alan
New crossover design
Originally Posted by alan-1-b
Hi Wayne,
Enquire Points (1) <---> (5) inclusive first, and Post the replies, and then I will discuss the Bass issues ... if there seems to be any sense in proceeding with this design.
Hi Alan,
The crossover configuration is optimum for the actual drivers in the baffle at 2m on and off axis.
Some adjustment to balance may be required due to the reflective characteristics of my living room, speaker location and my personal preference. So once I have listened to the result I can then make a judgment and changes can be made.
Regarding the mid-range damage due to the change of C7 from 24uF to 27uF – this will not cause damage to the mid-range driver. Calculating the difference in effective attenuation at the lower crossover frequency (500Hz) between the two capacitor values give a 0.54dB difference - this is barely audible at this frequency (about 1dB is audible) and will increase the mid-range diaphragm excursion only slightly i.e. about equivalent to increasing the volume on my amplifier by about the same amount. 27uF is used because it gives a flatter response anecholically.
I would be interested in your views regarding the bass section.
CheersWayne
About the midrange drivers:
Hi Wayne,
are the plots' measurements of YOUR particular MD500s, or from other samples ? _________________
Alan
Originally Posted by alan-1-b
Hi Wayne,
Are the plots' measurements of YOUR particular MD500s, or from other samples ?
Hi Alan,
They are mine.
Regards
Wayne
Hi Wayne ... and anyone else interested in "improving" the bass response.
The original inductors in the bass filter have significant DC resistances with regard to the 4 ohm DC resistance bass driver,
and which is only a little above 4 ohms in the upper bass region of its used bandwidth,
thus there is larger signal loss in the inductors than optimum,
BUT, remember that these bass drivers were used with lower efficiency mids and tweeters, especially the tweeters, thus for the original 66 those inductors do cause the bass drivers to match the output of the other drivers, and do cause a similar Transient Response in the lower mids to what was in the upper mids via the MF or MD mid-domes with the original capacitors there with their significant ESR.
The 66 is a balanced design in its particular type of sound.
It can be modernised in sound, and for some listeners that will be an improvement, thus:-
To significantly reduce losses in the bass filter's inductors the DC resistances need to be at least half the original values, and lower than half is better there.
As two 66 owners have posted DC resistance measurements in this thread for those inductors:-
1.5 ohms and 1.3 ohms,
thus such are likely close enough to what is there
{allowing for difficulty in measuring very low DC resistances, and errors in Multimeters}.
The 15AWG Jantzen P-core inductors that Wayne has in his notional improved filter have DC rest's of .255 ohm and .192 ohm according to the Jantzen specifications,
and that is good, BUT these are cored inductors ...
I have no doubt that Jantzen's technical measurements for this Permite 75 alloy are correct, and that this alloy allows better results than Ferrite for cores,
however, Air-core inductors will cause better result, and particually when the music is loud.
To get such low DCRs with air-core, 12AWG wire would have to be used,
and inductors using such will be very large physically, and very heavy,
however, the DCRs so not need to be quite that low, and with air-cores the sound will be sufficiently better that slightly larger DCRs will not be a hindrance.
With 14AWG wire it is possible for a 3.5mH inductor to have a DCR of .32 ohm,
and that is more than good enough.
The actual DCR with 14AWG for 3.5mH is dependant on the physical dimensions of the coil, because those dimensions determine how much wire is needed for 3.5mH.
For various options of sensibly chosen dimensions the DCR varies between .32ohm and .41 ohm,
and such inductors will be between 3" and 4.5" in diameter.
Jantzen have a 14AWG inductor of 3.4mH with a DCR of .488 ohm,
thus it is not wound to the optimum dimensions, but it is low enough in ohms to be better than the original Celestion part, and 3.4mH is close enough to 3.5 to not cause an audibly noticeable difference, and if it was noticed it would likely be better, because with lower DCR there does not need to be quite as large inductance in a Series connected inductor.
Alternately, if 3.5mH is wanted, the Jantzen air-core using 15AWG has DCR of .69 ohms - which is still less than half the DCR of the Celestion part.
Similar is the case with Jantzen air-cores for the 2.2mH.
Options are .385 ohms in 14AWG, and three with slightly larger DCRs,
and in 15AWG is 2.25mH {close enough} with .63 ohms.
I have to go now - I'll continue this next time.
Hi Alan,
Sorry about the format of the previous extracts, but I have tried to encompass most of your misgivings about my new crossover design.
I have been doing a lot of listening to various types of music.
The problem I have is one that you pointed out a long time ago, and that is that so much has been changed it is difficult to pinpoint the exact nature of the improvements in the sound they now produce. Also, it is difficult to determine where the sound is not as good. The one thing that is clear is they have a different sound from the original speakers I had. This of course is bearing in mind that I originally inherited the Realistic Super-Tweeters having been used to replace the original HF2000's.
The bass is much tighter and it's easier to follow bass lines. The new SEAS Tweeters sound good, and I have been pleasantly surprised about the improved clarity of the vocals, especially female vocals. I do have some issues regarding the midrange. I know that the designer, John, found there to be a dip in frequency reponse at 4,500 Hz on one of the midrange drivers. So he reduced the crossover point to 4000 Hz instead of the original 5000 Hz. I still need to speak to him regarding the possible impact this may be having on the midrange sound, as I believe this is now where the crossover to the tweeter takes place.
It's not always to determine which instruments/vocals etc. to listen out for between 4 - 5 KHz!
I know you have misgivings regarding the capacitor in the midrange circuit. This has now been changed to 27uF, which you believe to be too high. On my original crossovers this component was made up of a combination of 24uF & 6uF, creating a value of 30uF, which is higher still from what I have now.
I am also rather wresting with the issue of how much internal sound dampening I should use. As I previously said, with the use of an ABR it kind of makes the speaker design somewhere between an infinite baffle and a ported design.
I would really appreciate your additional views regarding the new midrange design and the change in crossover point from 5kHz to 4kHz, bearing in mind that one of the mid drivers had a large dip at 4.5kHz. I think I have already posted the various plots and graphs.
As I have previously said, this is not my final attempt at designing the crossover circuitry. John has given me specifications for future upgrades when I have the money to do so. I do however realise that you cannot base your entire design using testing measurements alone, and that it is as you have said, a compromise!.
I would also very much appreciate your views regarding sound dampening of the speaker's cabinets.
Best Regards
Wayne
Last edited:
resistors added to crossover
I've installed the 1 ohm, 1.5 ohm , 1.8 ohm, 2.7 ohm and 82 ohm resistors and put the 68uF caps with the highest values in the circuit parallel with the woofers as you suggested. I made no other changes.
I've had a quick listen with a few different CD's and FM broadcasts. The 66's now sound really good, very musical again. Very open. They seem like completely different speakers. The blandness in the sound evident previously is gone. Amazing result with what appears to be such minor changes in the crossover circuitry. Hard to believe in fact. Thanks Alan for your very expert advice.
With the limited listen I've had there is perhaps a little overbrightness (for want of a better term) in some material in the upper mid/treble??. However, as I've yet to fully evaluate them fully I can't be sure - will do so over the next month or so. 66's are very revealing. I'll also do some (amateur) freq response measurements after Christmas.
I'm now keen to fine tune them. I guess the next thing would be to replace the mid and tweeter caps. Would the Sonicap Gen I be suitable or the ClarityCap PX. The Sonicaps have suitable values: 24uF, 6uF (and 6.2uF), 4uF all with a 200VDC rating (4x caps are about $70 from Sonicraft with 20% off this month = $56). ClarityCap PX have 25uF, 6uF and 3.9uF (4x caps are about $40 from Madisound). I can see you can pay a small fortune for caps! I want a good step up from the Solens but don't want to go overboard.
DG
I've installed the 1 ohm, 1.5 ohm , 1.8 ohm, 2.7 ohm and 82 ohm resistors and put the 68uF caps with the highest values in the circuit parallel with the woofers as you suggested. I made no other changes.
I've had a quick listen with a few different CD's and FM broadcasts. The 66's now sound really good, very musical again. Very open. They seem like completely different speakers. The blandness in the sound evident previously is gone. Amazing result with what appears to be such minor changes in the crossover circuitry. Hard to believe in fact. Thanks Alan for your very expert advice.
With the limited listen I've had there is perhaps a little overbrightness (for want of a better term) in some material in the upper mid/treble??. However, as I've yet to fully evaluate them fully I can't be sure - will do so over the next month or so. 66's are very revealing. I'll also do some (amateur) freq response measurements after Christmas.
I'm now keen to fine tune them. I guess the next thing would be to replace the mid and tweeter caps. Would the Sonicap Gen I be suitable or the ClarityCap PX. The Sonicaps have suitable values: 24uF, 6uF (and 6.2uF), 4uF all with a 200VDC rating (4x caps are about $70 from Sonicraft with 20% off this month = $56). ClarityCap PX have 25uF, 6uF and 3.9uF (4x caps are about $40 from Madisound). I can see you can pay a small fortune for caps! I want a good step up from the Solens but don't want to go overboard.
DG
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.