Using Joachim Gerhard Filter Buffers as balanced crossover

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Further to an idea from EUVL regarding passive filter network followed by JFET follower,http://www.diyaudio.com/forums/pass-labs/225230-f5x-preamp-6.html#post3300675 I am seeking advice in using 4 of Joachim Gerhard Filter Buffers http://www.diyaudio.com/forums/grou...gerhard-filter-buffer-es9022.html#post3101908 as a balanced crossover. The intention is to use the up coming F5X preamp into this crossover feeding 2 fX5's.
The idea is to use the filter buffers for the High and Low pass filters (using 4 of each filter to get the LR and balanced output) ditching the original input filters and replacing with a 2nd order Sallen-Key as below. I settled on Q=.5 (Linkwitz-Riley) as the Butterworth filter calculations I tried always seem to oscillate according to Okawa electronic designs calculator.

Fc 2800Hz, R’s = 12.1K, C’s = 4.7nf

As I understand the outputs from the LR2 crossover filters are 180 degrees out of phase (high pass to low pass), would I be correct in that one of the speaker drivers needs the polarity reversed or would I reverse the either the low pass or high pass balanced outputs into one of the FX5’s?
Any help or views on the above would be appreciated.

Tony
 

Attachments

  • Original.png
    Original.png
    10 KB · Views: 838
  • High pass.png
    High pass.png
    8.5 KB · Views: 873
  • Low pass.png
    Low pass.png
    8.5 KB · Views: 832
Hi,

I'm using this kind of buffer for several years now in filters and it works very fine. Sonically it leaves OPamp graveyards far behind.
The combination of 2SK389/LSK389 and 2N/MMBF/SST4391 guarantees for low offset without the need of matching devices or dc blocking caps.
I use the SMD parts, thermally coupled by a cooling fin covering all 4 devices. Temperature drift is then no problem and the whol buffer builds very compact.
If more current or low load impedance stability is required the buffer may be ´boosted´ in a Sziklay-style fashion as I showed http://www.diyaudio.com/forums/analog-line-level/226099-preamp-buffers-simple-idea.html

jauu
Calvin
 
Last edited:
The idea is to use the filter buffers for the High and Low pass filters (using 4 of each filter to get the LR and balanced output) ditching the original input filters and replacing with a 2nd order Sallen-Key as below. I settled on Q=.5 (Linkwitz-Riley)
This is a description of an LR4 crossover.
It is 4pole roll off.
No two drivers I know will work well in a 2 way with an electrical
2nd order x/o at ~ 3KHz, active or passive with a bassmid zobel.

Can we have a comment or comments on whether an LR4 can work with conventional drivers that have a good working passband that extends well outside the LR4 pass band limits?
 
APK,

You need to consider if you are targeting a specific electrical response or acoustic response. The roll off of the drivers, enclosure (baffle step) etc will contribute to the acoustic response. Targeting an electrical response might not be the best way to develop a crossover design.


Otherwise as a fellow builder of JG filter-buffer ... neat idea and application of these boards!! I will follow with interest and have no specific advice to offer other than the general waffle above :)

Chris
 
Hi,

Yep, I assist that suggestion ;-)
With passive speakers, no one would expect good results from a Xover after the textbooks. With active Xovers though its seems rather common to design after textbook. In better designs one may find additional EQs to correct for driver anomalies. Unfortunately such designs tend to be OPamp graveyards. The sonic results of suchalike leave alot to wish for.
There´s basically no reason not to design active filters similar to passive technology with the drivers transfer function in mind, hence to perfom filter and equalizer-function at the same. Typically such filters place a lower number of parts and active devices into the signal path and sound much more musical/authentic.

jauu
Calvin
 
Hi Calvin & Hochopper
I understand what you are saying however with very limited knowledge I'm finding even this daunting, first looks simple however like everything else scratch the surface and all sorts of complications arise.
I think the initial plan is to follow the specific electrical response route and once I have achieved that and hopefully learning along the way,tackle what I am assuming is the more complicated path of the acoustic response.
 
From most the circuit's I have seen each stage is followed by a unity buffer which would mean using twice the number of JG Filter buffers (which isn't a problem) as indicated by Calvin in post #14. EUVL would seem to suggest in post #11 that this in not needed? the only circuits I found that don't have a buffer after each stage have some amplification as below.
As I have already said I'm fine with the practical side but c##p at theory
 

Attachments

  • High pass 1.PNG
    High pass 1.PNG
    29.1 KB · Views: 252
  • High pass 2.PNG
    High pass 2.PNG
    29.6 KB · Views: 248
Hi,

See #8, 2nd sentence! Each 2nd order SK filter stage has its Buffer. Since the Buffer is a low impedance output node, no additional Buffering towards the output node is required. Hence, a 4th order SK requires 2 Buffers. A 3rd buffer is not needed. Of course should the buffer circuit be capable to drive the attached filter components and output load. In case of the simple cascoded JFET buffer here, load impedance should not drop much below ~10kOhms. If more drive capability is needed, then use a beefier Buffer as for example shown in #5.

jauu
Calvin
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.