Custom Audio Amplifier Experiment

Since Peerless do not provide speaker grills with their drivers, got some discarded Polkaudio & JBL Infinity car speaker grills from a Car audio installer to cover the woofers.

6.5" is a standard size in car audio hence the grills were perfect match in terms of diameter size & mounting holes.
BudgetSpkr3_zpshxnzh301.jpg


Got pre-fabricated speaker boxes made of 0.5" (inch) thick plywood. It is covered with black polythene sheet.
Dimensions are H 12.5" x W 9" x D 8.5". Filled the rear with 1" thick glasswool. Fixed ports of 1" ID & 2" Length.
BudgetSpkr4_zpsx0ptys05.jpg
 
Last edited:
Decided to use JBL covers instead of Polkaudio since I couldn't get the center most net caps for Polkaudio & moreover JBL has a chiseled look. Took a bit of cutting internal plastic tabs to make it seam with the woofer frame.

To mount the tweeters, I used chromed nuts with finisher covers which are used usually to mount mirrors.

The passive crossover PCB was screwed onto the lower plank below the woofer.

This is the finished product - budget Bookshelf speaker -
BudgetSpkr5_zpspmrjcbl9.jpg


& Yes I will remove those JBL Infinity proprietary brand tabs before I get sued for copyright infringement !!!
 
Last edited:
Thin metal frame border speakers do not look that great
Peerless 6.5" S16NI Peerless - 6" Coated Paper Cone Woofer - S16NI

as compared to thick plastic frame border speakers
Peerless 5.25" M13NH Peerless - 5.25" Coated Paper Cone Woofer - M13NH

When it comes to the sound stage, I like how these speakers respond to music. There is a suppression of frequency band of around 4kHz-7kHz negating the use of graphic equaliser. However it would not be ideal for a home theater setup where the mid/vocal range is important for listening to movie dialogues.

Am contemplating making compact speakers using full range 4" Dayton reference series RS-100 for the future.
 
LM3886 modification: -

I came across this modification accidentally. Was wondering why the robust LM3886 chip as compared to TDA7294 (earlier test) was not delivering the deep bass response at the same VA levels +-35v 2A.

On the Wintek LM3886 board noticed twin 47uf capacitors soldered with their +ve legs tied together protruding out from the board.

Lm3886mod_zpsvodi8uh8.jpg


A glance across TI's datasheet understood that these are negative feedback capacitors which limit the frequency response.

TIamp-LM3886_zpsvak592xi.jpg


The non-polar capacitor labelled Ci (with an *) with a value of 22uf. To get the closest value for a non-polar capacitor the manufacturer joined 2x 47uf polar capacitors in series with their +ve side facing each other.

Hence effective value CapSeriesTOTAL
= 1/ ((1/C1) + (1/C2))
= 1/ ((1/47) + (1/47))
= 23.5uf

This value is close to the TI recommended 22uf. Came across few LM3886 circuits where the response was reported better with non-polar 47uf feedback capacitor.
Tried it (replaced twin orange caps with single grey) & the chip sounds much better. Clean highs & deep bass. The downside is the chip runs slightly hotter which I personally feel due to the increased output swing.

The LM3886 runs quite cool with +-12v power supply. Might experiment with low-noise 2x 12v 3A SMPS & observe the results.
 
Last edited:
Do tell the results with +-12v supply and smps, i have two chipamps board ready for operation just want to see if 12v supplies work and how much power would they deliver
 
LM3886 modification: -

I came across this modification accidentally. Was wondering why the robust LM3886 chip as compared to TDA7294 (earlier test) was not delivering the deep bass response at the same VA levels +-35v 2A.

On the Wintek LM3886 board noticed twin 47uf capacitors soldered with their +ve legs tied together protruding out from the board.

Lm3886mod_zpsvodi8uh8.jpg


A glance across TI's datasheet understood that these are negative feedback capacitors which limit the frequency response.

TIamp-LM3886_zpsvak592xi.jpg


The non-polar capacitor labelled Ci (with an *) with a value of 22uf. To get the closest value for a non-polar capacitor the manufacturer joined 2x 47uf polar capacitors in series with their +ve side facing each other.

Hence effective value CapSeriesTOTAL
= 1/ ((1/C1) + (1/C2))
= 1/ ((1/47) + (1/47))
= 23.5uf

This value is close to the TI recommended 22uf. Came across few LM3886 circuits where the response was reported better with non-polar 47uf feedback capacitor.
Tried it (replaced twin orange caps with single grey) & the chip sounds much better. Clean highs & deep bass. The downside is the chip runs slightly hotter which I personally feel due to the increased output swing.

The LM3886 runs quite cool with +-12v power supply. Might experiment with low-noise 2x 12v 3A SMPS & observe the results.

The twin 47uf is a very low value for that place. you need atleast 100uf of capacitance there, thats why you feel lack of bass. And you dont need a bipolar there, polarised will do a fine job, you do need a very good quality cap there, silmics, muse kz, panny fms, fcs are the usual choices.
 
Ri and Ci are in negative feedback loop.
With 1K and 23.5uF ( 2x 47uF in series) the low frequency roll off ( -3dB point ) is about 7Hz. This should be OK and should not result in loss of bass.
I notice no input capacitor. If your source has no dc then it's OK. If it does then you need to add a capacitor there to prevent dc at the output.

This one is critical. It can roll off bass early if too small and affect the sound higher up in frequency , depending on it's type. Usually polypropylene film caps are used. An electrolytic cap is also possible ( non polar or two caps back to back ). Some polyesters are also OK.

If you use only a single electrolytic caps at Ci and at the input , then check the circuit to determine the dc status at that node. Align the capacitor accordingly ( +/-).
 
Ri and Ci are in negative feedback loop.
With 1K and 23.5uF ( 2x 47uF in series) the low frequency roll off ( -3dB point ) is about 7Hz. This should be OK and should not result in loss of bass.
I notice no input capacitor. If your source has no dc then it's OK. If it does then you need to add a capacitor there to prevent dc at the output.

The loss of bass will be audible with a 7hz filter, unless you are using satellites. Your speakers will also have a rolloff. You need to be about a decade clear off of that. If your speakers have -3db at 30hz, your nfb should rolloff at 3hz.

If an input cap is used, then nfb rolloff should be lower than that. The rolloff of the amp should be decided by the input cap network, not by the feedback loop.

It is usually recommended to have either ac coupling or dc coupling at both input and fb loop, mixing isnt supposed to be a good idea.
 
Another project - decided to make a monoblock TDA7498MV with SMPS, DC Filter circuit & 6db crossover.

Tried with both 12v & 24v SMPS. Found 24v to be better in vocal \ treble clarity.
Also better at high volumes at reducing speaker 'pop' \ distortion sound. Speakers used were Dayton ND91-4 & ND20FB-4.



First picture is the motherboard to which the TDA7298MV & SMPS pre-made boards are mounted.
VF85Op0.jpg


Second is the finished monoblock with plastic spacers (to prevent any shorts).
qffHGsH.jpg
 
The reason why I pursued TDA7498MV is when I compared it to Mono bridged TPA3116D2 & I found the it (TDA7498MV) to be better in bass & high frequency clarity albeit slightly higher noise.

Tested with 19v HP laptop adapter & Dayton RS100 with dual magnet bullet tweeters.

Left red board TDA7498MV - Right blue board mono TPA3116D2
Qm3gp9x.jpg


As a follow-up on the Class AB project last year, decided to make Active bookshelf speakers. Narrowed down to

Amplifier kit - parallel LM3886 from eightaudio.
Speakers - Dayton RS180-4 Midrange & Pyramid TW18 Tweeter.

Speaker enclosure fabricated by diyaudiocart.
fGPO2zZ.jpg


For power supply tried first with 2x 12v 5A separate SMPS. Internal they were modified to reduce DC line noise. Works fine individually (mono)but as soon soon as the next speaker (stereo) is added it created a slight high frequency buzz - ground loop issues. The tweeter being a highly sensitive (high SPL 100db+) added to the woes. Fixing a capacitor-resistor ground loop isolator could not solve the issue.
YDpd2LN.jpg



Then removed it & fixed toroidal 12-0-12 160VA transformer with 6800uf x4 per rail capacitor bank. Noise issue resolved. One of the major reasons running low voltage & higher amps is LM3886 runs much cooler.
XnHJHfh.jpg


Got the passive crossovers from eightaudio.

First had 12db Butterworth crossover at 2500Hz. Then changed it to 12db Linkwitz Riley crossover at 3100Hz.
Left Linkwitz Riley - smaller capacitors & bigger coils vs. Right Butterworth - bigger capacitor & smaller coils
GAmgGay.jpg

The tweeter SPL100+ is so high as compared to the midrange SPL89 that I had to use 15? resistors before the high frequency crossover to equalize the sound.

Tweeter polarity is reversed compensating for phase shift.
TSsfIX3.jpg


Rear of the Speaker enclosure has audio RCA connector (MX Gold type), 230v AC EMI line filter, 2x fuse holders & 2x LED for monitoring rail voltage.
eNuurAb.jpg


Finished bullet speaker project without \ with grill. The front gill guide fitment is perfect such that right grill fits on left speaker & vice versa. Surface is polished and lacquer coated. Ports are Jantzen-audio 2".

Diy5Pjw.jpg


The wooden enclosures are double walled in the front to compensate for the recess cut for speakers to be flush mount to the surface. With that and all the contraption inside to lift the two speakers one should be a regular at the gym ! - so heavy.
 
Last edited by a moderator:
Used a sheet of yellow glasswool between the speakers & rear wall. It reduces the boominess & give bass a tighter punch. Outer dimensions are : 8.5"W x 18.25"H x 9.5"D = 1474 cubic inches total volume. Wood is MDF.

Weighed it today - 12.3 Kgs per box ! - with everything inside.
728ICTU.jpg


Specification sheets of the drivers -

Dayton RS180-4
dzAebJC.jpg


Pyramid TW18-4
no7DLcw.jpg

4jbAZ5D.jpg


Modulating Mid-range & Taming Tweeter :
Initially I had the RS180 (SPL 89dB) crossed below 2.5kHz. I found the Pyramid TW18 (SPL 100dB+ - specs say 96dB but its way higher) overpowering even after using 15? 5W resistors before the passive high pass network giving attenuation of 13.5dB. Between 2kHz - 8kHz the tweeter is so powerful that off-axis bass was literally absent. What impressed me about the tweeter was its performance beyond 10kHz - exceptional crisp treble.

To reduce this effect later I raised the crossover point to 3.1kHz. RS180 with the solid fixed metal cone (non-moving fixed onto frame) is known to exhibit a ringing\beaming effect if crossed above 2.5kHz 12dB/octave. Hence made an active high-pass pre-amp filter above 70Hz. This worked out for the mid-range (70Hz - 3.1kHz). Am planning to add a subwoofer later to take of lower <80Hz frequencies. However still the tweeter was still giving me grief - the 'sibilance' in the instruments\vocals especially artists who use autotune was hitting the ears.

For a solution to tame the tweeter decided to implement a parallel notch filter.
 
Last edited by a moderator:
Last edited:
I added the parallel notch filter few days back.
9sv4aGy.jpg


All I can say I am amazed by the transformation in the tweeter's response ! Till about 8kHz where the peaks were prominent its absent now. The tweeter's mid-frequency response is closely in-line with the midrange's response.
 
Last edited:
Found time to finish a Class D project. Boxes were fabricated by diyaudiocart.

Mini bookshelf active speakers. Class D TDA7498MV with 24V 4A SMPS monoblocks mounted on custom power filter & crossover circuit board. Drivers are Dayton ND91-4 & ND20FB-4. Crossover at 6KHz 6dB/oct.
 

Attachments

  • ClassD1.jpg
    ClassD1.jpg
    79.7 KB · Views: 14
  • ClassD2.jpg
    ClassD2.jpg
    44.1 KB · Views: 15
  • ClassD3.jpg
    ClassD3.jpg
    95.2 KB · Views: 16
  • ClassD4.jpg
    ClassD4.jpg
    82.3 KB · Views: 13
  • ClassD5.jpg
    ClassD5.jpg
    72.8 KB · Views: 10
Wharfedale Linton Heritage Speakers in Walnut finish at a Special Offer Price. BUY now before the price increase.
Back
Top