Distortion Tests - HifimeDIY T2 / LJM L20D / SURE TDA7498

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

For yuanjing board, I reckon the power capacitor(due to 35v rating) & speaker output inductor needed to upgrade accommodate TDA8950 more voltage swing.

TDA8950 VDD is +-41V to produce 300W (bridge).

The question playing in my mind is , Can a class-d chip amp better & more powerful than discrete amp.

I've a TAS5630 amp with me, this chip works great on low impedance speaker but it's not powerful enough for 8ohm speaker. It can only do parallel bridge for 2-4ohm speaker.

Another candidate is NXP TDA8953 / TDA8954 which claim 400W ++ in bridge mode.
This chip is hard to get and no measurement/review by third party.

I always feel NXP over exaggerate the specs.

And lastly, thank you for sharing your experience with us.
 
Lithoc,
The physical size of the speaker output inductor is primarily determined by the maximum average current it has to carry. You are correct in being suspicious of the Yuanjing inductor. They use an inductor somewhat similar to the Sagami 7313NC-220 part. This is rated to handle around 4A max, or 4 x 4 x 8 watts of power into an 8 ohm speaker. A good replacement, which however will not fit directly into the TDA8920 board, are the 1D14A-220M from Mouser Electronics (inter alia) which will handle 7.5 amps max (7.5 x 7.5 x 8 Watts max). The 1D14A are actually about the same size as the original 7313NC part, but the pins will not solder directly in unless you bend the leads a bit...

Power capacitor voltage ratings need to be watched, and it is best to buy a "Low ESR" type, with a lower internal resistance.

As for the best high power amp I have seen to this point, please see the following post :) The L20D can be scaled in power rating and voltage from the basic board, a modder's dream, in fact. Its inductor, a 1D17A-220M (equiv) will handle around 17 amps :)
.
 
After being prompted by Saturnus, above, I decided to buy a more sensitive 24bit analog to digital converter for my Distortion tester setup.

It is very hard to measure distortions around -80dB, which are already 10dB down on full scale (instrument overhead), and at the 1watt level, 40dB down from that. The 16 bit D-A converters cannot handle the dynamic range, whereas my new 24bit converter can resolve down to a noise floor below -120dBm.

So I measured jlw_jlw's L20D again, the low cost IRS2092+IRFI4020 design. I increased the supply voltage capacitors to 100V rating, 330 uF, otherwise the boards are unmodded, exactly as received from Zoe on Ebay.

The two-tone Intermodulation distortion at 35 Watts into 8 ohms, 1KHz, (attachment 1), is as clean as one could expect from a high quality switching amplifier (compare it with the plot from the TDA7498 which I posted in another thread earlier today).

With the higher resolution A-D converter, the distortion vs power curve resolves noise-free by comparison with the curves I posted to begin this discussion thread. You can notice the quantization dither changes from step to step as the input level was incremented to move us across the curve from low to high powers. The supply was a real-world 100W toroid supply with 88,000uF total capacitance, giving 50-0-50 no-load. I used an 8 ohm load for these tests.

I have mentioned the quantization error, when looking at a signal with this level of precision. In order to investigate further, I displayed the 2nd and third harmonic distortions only (red=odd order, blue=even order). It is interesting how the IR quantizer is dithering the second harmonic distortions more than the higher (odd order) components (odd order usually created by clipping, such as at the 100W+ power levels here).

When measuring a switching amplifier with this amount of dynamic range I could obtain a smooth curve by spline-smoothing the data points, or by applying an A-weighting filter (or indeed some other weighting) prior to measurement. I don't do that sort of thing, so my curves will always show exactly what is present, rather than smooth lines :)

Anyway, that's my summary of the L20D, IMO a superb amplifier implementation, clean, stable, and simple. It is also cheap and easy to replace the power chip. I would like onboard speaker protection, as well as IR's overcurrent/overvoltage protection circuits to be added, but as a high-end switched amplifier, I am very happy with this one. Two are going into my new MinDSP-based electronic-xover amplifier system (the one with Linkwitz Pluto-like speakers) :)
.
ps: -60dB = 0.1% while -80dB = 0.01% distortion
 

Attachments

  • IM_L20D_17Vrms_8ohm.gif
    IM_L20D_17Vrms_8ohm.gif
    16.3 KB · Views: 2,879
  • Pdist_L20D_8ohm_50-0-50Vnl_THD.gif
    Pdist_L20D_8ohm_50-0-50Vnl_THD.gif
    17 KB · Views: 2,860
  • Pdist_L20D_8ohm_50-0-50Vnl_twothree.gif
    Pdist_L20D_8ohm_50-0-50Vnl_twothree.gif
    16.2 KB · Views: 2,823
Last edited:
UPDATE - two minor errors in the post above. Obviously it is a LACK of dither on the even order delta-sigma error components which would lead to variation in the quantization distortion from step to step. Secondly, I wrote the designer's name incorrectly, the author of the L20D design is DIYaudio member ljm_ljm. More info on the amplifier is in this thread: http://www.diyaudio.com/forums/class-d/191739-my-design-l20d-irs2092-irfi4020h-200w8r.html

Incidentally, I did experiment with a number of inductors, including the Sagami 7313NC, 7B12 series, as well as ICE Components 1D14A-220M and 1D17A-220M and I can confirm figure 31 of the iraudamp7s.pdf document showing the THD with different inductors. The one which ljm_ljm chose is indeed the one with lowest distortion at power - the 1D17A-220M.
.
 
Last edited:
I am sorry, I don't feel comfortable quoting noise levels on amplifiers as good as the L20D. There are too many variables, weighting/no-weighting, etc. I did put a small speaker on an L20D module, and with my ear in the speaker I can just hear a high-harmonic 60Hz 'buzz' from my mains power supply and/or random signal pickup / ground loops / on my workbench -- along with a predominantly low frequency pink noise hiss which is barely audible.

As far as I am concerned, it has no noise, nothing you can hear 1 foot from a speaker in a 30dB listening room. But your measuring preferences may vary :)
.
 
True, the widerange speakers I stuck to my ear are 84dB/W, Peerless 830983, But they cover from 40Hz to 20K, as long as you don't want a lot of volume... and as long as you use a MiniDSP to equalize them :)

These days I myself would select a higher power, highly efficient amp (like the L20D) and a wider range, very low distortion, speaker :)

Still, 113dB/W is a pretty impressive achievement :)
.
 

Attachments

  • Tiny_Speakers.jpg
    Tiny_Speakers.jpg
    66.3 KB · Views: 2,732
Thanks for that. Everything's a little different with 113dB/W drivers :)

No, it isn't. trevmar's speakers with 84dB/W/m sensitive speakers can quite adequately and easily illustrate the no sound noise of the amp you'll experience in your listening position. The 29 dB difference in sensitivity, let's just say 30 dB to keep it simple, is just 5 doublings or halvings in distance.

Let's say your listening position is 4 meters from your speakers, then the sound output is 101dB@1W output, so you can say your speaker is 101dB/W/4m. And if trevmar is holding his speaker 12.5 cm from his ear, that means the sensitivity is 102dB@1W, or 102dB/W/0.125m.
 
Saturnus,

Thanks for your explanation. Perhaps my choice of the word 'different' was not good.

Unfortunately, when using compression drivers nearfield (~1.5m in my case), it really is important to get low noise amps. The Panasonic class D receiver I have been using is too noisy.

:)
 
Saturnus,

Thanks for your explanation. Perhaps my choice of the word 'different' was not good.

Unfortunately, when using compression drivers nearfield (~1.5m in my case), it really is important to get low noise amps. The Panasonic class D receiver I have been using is too noisy.
:)

And trevmar's explanation will strongly suggest there will be no noise at your listening position. He said he had to move the speakers closer than 5cm from his ear to hear anything at all. 5cm for his speakers equals 1.5m for your speakers.
 
The Panasonic class D receiver I have been using is too noisy
Usually the preamplifier chain determines the overall noise level of a receiver. If the preamplifier is all-digital, then the preamplifier noise contribution drops to very low levels. That is why I chose to use a Mini-DSP for my "tone-controls" and digital volume control (the MiniDSP has a DC Potentiometer / ADC to set the output level). It is somewhat weird to be able to set up precise tonal compensation via my netbook, but it is also very effective. I don't have to do it very often, as flat sources and flat speakers mean most material sounds excellent without any shaping... If you use optical audio distribution (SPDIF) then you can buy digital switches to select among multiple inputs. I have been using an HDMI switch with SPDIF output, but am now switching to one with analog outputs (which handles AC3-WAV conversion internally).

ps: This is the switcher I use, it has both Toslink SPDIF and Analog outputs:
4X1 HDMI 1.3b Certified Switcher w/ Toslink & Digital Coaxial Port (Rev.2) w/ 3D support.
I use this device to convert my old DVD player to HDMI:
Component (YPbPr) & S/PDIF Digital Coax/Optical Toslink Audio to HDMI Converter
.
 
Last edited:
Desibros,
I don't know what you read on the Ebay page, I always work out the connections by looking at the circuit board, and seeing where the contacts go. Looking from the top of the board, the order of pins the header block, from the end where is printed the "Voltage DC +-70V Output Power 250W 8R" is -V, Ground (zero for power supply and one side of speaker), Speaker active side, and +V

I hope that helps. If in doubt, you can check where the + and - go to the main bypass capacitors...
.
 
Desibros,
I don't know what you read on the Ebay page, I always work out the connections by looking at the circuit board, and seeing where the contacts go. Looking from the top of the board, the order of pins the header block, from the end where is printed the "Voltage DC +-70V Output Power 250W 8R" is -V, Ground (zero for power supply and one side of speaker), Speaker active side, and +V

I hope that helps. If in doubt, you can check where the + and - go to the main bypass capacitors...
.

Hello! :) thank you a lot the suggest was usefull! :)
Bye!!
max
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.