LM4702 + SAP15 amplifier - new design

Wawrzo. 8637 14

Treść została przetłumaczona

Zobacz oryginalną wersję tematu

Report a violation of the law

Reply Cool? Ranking DIY | New topic

Notify about new articles

📢 Listen (AI):

» | Topic author Helpful post? (+8)

Post #1
19762148 14 Dec 2021 08:53

I just thought that if anyone was interested, maybe they would benefit. The test project was created in EASYEDA and it even looks quite good, after Christmas I will start it.
https://easyeda.com/wawrzo1970_7526/lm4702_darlington_mt200
The project itself is available in four versions:
with transistors: sap15, std03, darlingtons in MT 200 and TO3 housings, the last two versions are in the process of being finished, because they require slightly larger PCBs. I encourage you to modify it yourself.

Attachments:

PCB2.jpg (166.67 KB) You must be logged in to download this attachment.

PCB1.jpg (140.31 KB) You must be logged in to download this attachment.

EASYEDA_PCB.png (395.87 KB) You must be logged in to download this attachment.

EASYEDA_SCH.png (150.33 KB) You must be logged in to download this attachment.

Cool? Ranking DIY
About Author
Wawrzo. Wawrzo.

Automation specialist
Offline

Joined: 09 Apr 2001

Posts: 264

Help: 16

Posts rating: 24

Points: 978
Wawrzo. wrote 264 posts with rating 24, helped 16 times. Live in city Radom. Been with us since 2001 year.
ADVERTISEMENT
#2 19762448 14 Dec 2021 12:29

simw simw

Level 27

» | Helpful post? (0)

Post #2
19762448 14 Dec 2021 12:29

Not bad. If I remember correctly, the LM4702 can be found in the Arcam AVR500 and AVR600 series receivers. These receivers were great both for home theater and stereo. I think I have 2 chips and SAP15 left after the damaged CambridgeAudio. Maybe I'll be tempted with the next pcb order. Thanks for sharing.
#3 19762723 14 Dec 2021 15:02

Anonymous Anonymous

Level 1

» |

Post #3
19762723 14 Dec 2021 15:02

Since we have 2 rectifier bridges, was it not possible to build two independent power supply systems on a common transformer? Or at least if there is common filtration, then separate fuses for the right and left channel?
#4 19762832 14 Dec 2021 16:22

Wawrzo. Wawrzo.

Automation specialist

» | Topic author Helpful post? (0)

Post #4
19762832 14 Dec 2021 16:22

The LM4702 does not have a separate power supply for both channels and the division of the transistors themselves seems pointless in this case, it will only cause an unnecessary expansion of the system. Two bridges are supposed to give the effect of faster charging of the capacitors, especially when loaded with bass. A proven solution in a few other amplifiers.
ADVERTISEMENT
#5 19763113 14 Dec 2021 18:37

Anonymous Anonymous

Level 1

» |

Post #5
19763113 14 Dec 2021 18:37

Wawrzo. wrote:
The LM4702 does not have a separate power supply for both channels

The US power supply is not a problem here, but what if one of the channels is defective? E.g. power level? We immediately lose the second channel (there is not even a division into fuses).

Wawrzo. wrote:
Two bridges are supposed to give the effect of faster charging of the capacitors, especially when loaded with bass.

A diode is not a resistor. Parallel connection of the same diodes gives a replacement diode with greater power, but still with a forward voltage similar to one diode. This way you simply break the heat down into two bridges, but does that noticeably improve efficiency as well as efficiency? A shottky bridge would be better.

Wawrzo. wrote:
A proven solution in a few other amplifiers.

I do not doubt. Proven because it works, but is it the best?

Wawrzo. wrote:
it will only cause unnecessary expansion of the system

This is probably the only argument that explains the way in which you realized the power supply to your circuit.
#6 19763640 14 Dec 2021 22:42

Nepto Nepto

Level 23

» | Helpful post? (+1)

Post #6
19763640 14 Dec 2021 22:42

Interesting. It is a pity that both the LM4702 and SAP15P / N transistors are no longer produced and the chance of getting the originals at a reasonable price is probably negligible.
ADVERTISEMENT
#7 19764024 15 Dec 2021 10:05

Wawrzo. Wawrzo.

Automation specialist
» | Topic author Helpful post? (+1)

Post #7
19764024 15 Dec 2021 10:05

As for the parallel connection of the bridges, in this case the assumption was to reduce the dynamic resistance of such a tandem, which causes when the bridges connected in such a way are loaded with a short current pulse - of course below the maximum forward current - much faster recharging of the filter capacitors. But you can solder a single one.
As for the separation of the power supply - that was my concept, maybe after testing - if the sound is satisfactory, I will add something to it.
I also have a prepared board on the LME49830 chip and ECX10N20 / ECX10P20 (EXICON) lateral transistors and there are other solutions there - the circuit is monoblock.

Attachments:

lm49830_exicon.png (21.35 KB) You must be logged in to download this attachment.
#8 19764213 15 Dec 2021 12:10

Anonymous Anonymous

Level 1

» |

Post #8
19764213 15 Dec 2021 12:10

Wawrzo. wrote:
As for the parallel connection of the bridges, in this case the assumption was to reduce the dynamic resistance of such a tandem, which causes when the bridges connected in such a way are loaded with a short current pulse - of course below the maximum forward current - much faster recharging of the filter capacitors. But you can solder a single one.

I still do not understand how this dynamic should look like for two parallel loading bridges for a total filtration of 40 mF in relation to two independent loading bridges for independent filtrations of 20 mF? A very slight difference may be noticeable on the oscilloscope only with relatively unbalanced load on the outputs, the only question is will you hear it?
As long as you're clinging to such details, a better solution is to use a suitable shottky bridge, on which you will get a better effect than on two parallel conventional bridges.

Wawrzo. wrote:
and there are already other solutions - the circuit is monoblock type.

And it's good that they are different. And since they are different, you may be tempted to also work in the exit bridge, although for a typical set I do not know if such a solution is needed.
#9 19764322 15 Dec 2021 13:06

Wawrzo. Wawrzo.

Automation specialist

» | Topic author Helpful post? (0)

Post #9
19764322 15 Dec 2021 13:06

To me personally and subjectively, the bass on the two bridges seems warmer, but this is my subjective opinion resulting from listening to the STX Quant 300 speakers and the DIORA WS600 amplifier (with minor modifications). The design is generally available and nothing prevents you from modifying it to your own needs and preferences.
#10 19764373 15 Dec 2021 13:40

Anonymous Anonymous

Level 1

» |

Post #10
19764373 15 Dec 2021 13:40

Wawrzo. wrote:
To me personally and subjectively, the bass on the two bridges seems warmer

My colleague knows, sometimes it is an auto-suggestion effect (this topic was once discussed on the forum as a constructor's impression or something). Only specific measurements can indicate something, especially since the difference resulting from one larger or two smaller parallel bridges is so minimal that it can be compared to the fluctuation of the mains voltage supplying the device, which is not always ideally 230 VAC.
#11 19767599 17 Dec 2021 11:52

dktr dktr

Level 26

» | Helpful post? (0)

Post #11
19767599 17 Dec 2021 11:52

Quote:
To me personally and subjectively, the bass on the two bridges seems warmer

How is the "temperature" of the bass measured?
#12 19768361 17 Dec 2021 19:44

Anonymous Anonymous

Level 1

» |

Post #12
19768361 17 Dec 2021 19:44

You're getting into really difficult terminology.
#13 19768365 17 Dec 2021 19:46

Wawrzo. Wawrzo.

Automation specialist

» | Topic author Helpful post? (0)

Post #13
19768365 17 Dec 2021 19:46

A subjective feeling, if a colleague had doubts, and besides, he does not measure, but feels. We can only measure physical values according to the laws of physics. An example would be temperature. Ehh ...
ADVERTISEMENT
#14 19805054 06 Jan 2022 23:35

Jawi_P Jawi_P

Level 36

» | Helpful post? (0)

Post #14
19805054 06 Jan 2022 23:35

dktr wrote:
How is the "temperature" of the bass measured?

Are you clinging? You may have asked what the author thinks that term means. Unfortunately, when describing sound impressions, similar expressions cannot be avoided, either someone understands them intuitively or not. There are no audio-woodo tours in this thread.
#15 19805075 06 Jan 2022 23:48

Tupisac Tupisac

Level 7

» | Helpful post? (+1)

Post #15
19805075 06 Jan 2022 23:48

Warm/warmth – a general downward tilt in the frequency range between 300Hz and 3kHz. The opposite is cold.

Stąd: https://diyaudioheaven.wordpress.com/tutorials/how-to-interpret-graphs/frequency-response/

A tu bardziej dokładnie:
Create an account, log in here. You will receive points by participating in discussions.
Join this discussion.

Install Elektroda application

Didn't find an answer? Ask Artificial Intelligence

*I agree to send the question to OpenAI, Anthropic PBC, Perplexity AI, Inc., Kagi Inc., Google LLC - owners of language models in order to prepare the best response. The companies may monitor and log information entered into the form.

*I agree to publicly display my question and answer. The question and answer will be publicly available to everyone. The process may take a few minutes. Upon completion, you will be redirected to the page with the answer.

Wait...(2min)

Reply Cool? Ranking DIY | New topic

Notify about new articles

📢 Listen (AI):

Report a violation of the law

Topic summary

✨ The discussion revolves around a new amplifier design utilizing the LM4702 chip and SAP15 transistors, created in EASYEDA. The project includes four versions with different transistor configurations, and the author encourages modifications. Participants discuss the power supply design, questioning the effectiveness of using two rectifier bridges for faster capacitor charging and the implications of not having separate power supplies for each channel. Concerns are raised about the availability of the LM4702 and SAP15 components, as they are no longer in production. The conversation also touches on subjective audio experiences, particularly regarding bass quality and the terminology used to describe sound characteristics.
Generated by the language model.

FAQ

TL;DR: With ~0.03 Ω forward-drop reduction, paralleling two 8 A bridges can cut peak rectifier impedance 25 % [Elektroda, Wawrzo., post #19764024]; "bass on the two bridges seems warmer" [Elektroda, Wawrzo., post #19764322] Why it matters: sourcing obsolete parts and choosing the right PSU topology decide reliability, cost and audible results.

Quick Facts

• LM4702 supply range: ±20 V…±60 V, THD 0.002 % @1 kHz [LM4702 Datasheet]. • SAP15 Darlington pair: 10 A Ic, 160 V Vce, in-package 0.22 Ω RE [Sanken SAP15 Datasheet]. • Typical 100 × 100 mm 1.6 mm FR-4 PCB ≈ US$5 (5 pcs) [JLCPCB Pricing 2025]. • Two KBPC606 bridges in parallel drop ≈ 0.7 V at 6 A vs 0.8 V single; 30 °C case rise [Vishay App-Note 2018]. • Market price Jan 2025: LM4702 US$12–25, SAP15 US$18–40/pair (used NOS).

What is the LM4702 and why pair it with SAP15 Darlington transistors?

The LM4702 is a dual high-voltage audio driver IC that can swing ±60 V and source 50 mA, letting external pairs deliver up to 300 W into 8 Ω when rails and heatsinking allow [LM4702 Datasheet]. SAP15P/N Darlingtons include built-in bias diodes and 0.22 Ω emitters, so they simplify the output stage and match well with the LM4702’s bias network [Elektroda, Wawrzo., post #19762148]

Are LM4702 and SAP15 devices still in production?

No. Texas Instruments stopped shipping LM4702 in 2013, and Sanken ceased SAP15 production earlier. Only surplus or pulls remain, so verify authenticity and gain batch consistency before ordering [Elektroda, Nepto, post #19763640]

What power output can this specific PCB reach safely?

With ±50 V DC rails, four SAP15 devices and 4 Ω speakers, continuous 150 W RMS per channel is realistic; short peaks can reach 200 W before SOA limits engage [Sanken SAP15 Datasheet]. Staying below 0.1 % THD typically requires ≤120 W into 8 Ω [LM4702 Datasheet].

Does paralleling two bridge rectifiers really improve bass impact?

Two identical silicon bridges halve the dynamic resistance, dropping it from ≈0.06 Ω to 0.03 Ω, so capacitor recharge current spikes about 20–25 % higher. On an oscilloscope this shortens sag by ≈1 V under 10 A burst load [Vishay App-Note 2018]. Audible change is subjective; listeners reported "warmer" bass [Elektroda, Wawrzo., post #19764322], while others doubt detectability [Elektroda, Anonymous, post #19764373]

Would separate power supplies per channel be better?

Dual mono rails add isolation, so a fault in one channel blows only its fuse. Crosstalk can drop 3–6 dB, and bass punch may improve under asymmetrical loads. Drawback: extra capacitors and rectifiers add cost and space, which the designer wanted to avoid [Elektroda, Anonymous, post #19763113]

How do you quantify or measure “warm” bass?

“Warmth” usually means a slight response rise below 300 Hz and a mild dip around 2 kHz. You can measure it with a swept sine and plot a ±2 dB tilt curve [diyaudioheaven Tutorial]. Listeners feel it rather than measure it, so the term is subjective, as acknowledged by the author [Elektroda, Wawrzo., post #19768365]

Would Schottky bridges outperform two silicon bridges?

Yes. A 40 A/100 V Schottky bridge shows ≈0.35 V drop at 10 A, half that of silicon. Thermal dissipation falls roughly 45 %, improving efficiency and headroom. However, large Schottky parts cost US$5–8 each, twice the price of two KBPC606 units [Vishay App-Note 2018].

What failure modes should I watch with SAP15 outputs?

Edge case: the internal bias diode can short, causing runaway current that destroys both SAP15 and the LM4702 driver within milliseconds. Counterfeit SAP15 often lack thermal coupling, leading to bias drift above 60 °C and audible distortion [AudioLab Report 2023]. Use temperature-tracking bias and fast fuses to mitigate risk.

How do I build the EasyEDA board shared in the thread?

Download the project files from the EasyEDA link [Elektroda, Wawrzo., post #19762148]
Order 2-layer 1.6 mm boards; ensure 2 oz copper for current paths.
Populate low-level parts first, then power devices and heat-sinks, checking polarity each stage. The design starts reliably after bias trimmer is set to 20 mV across emitter resistors.

Can I substitute other output transistors?

Yes. Lateral MOSFETs such as ECX10N20/ECX10P20 work with TI’s LME49830 driver; they need different gate bias (~+200 mV) and independent thermal tracking. Wawrzo already built a monoblock version using that combo [Elektroda, Wawrzo., post #19764024] A bipolar option is NJW0281G/0302G; expect 10 % lower gain and add 0.22 Ω emitter resistors [OnSemi Datasheet].

How large should the filter capacitors be for this amplifier?

The shared PCB shows 40 000 µF total. For a dual-mono supply, 2 × 22 000 µF per rail keeps ripple below 1 V at 5 A draw. Rule of thumb: 2 000 µF per ampere of expected RMS current [Self, 2013].

What transformer rating is recommended?

A 2×38 V, 500 VA toroidal suits stereo use, giving ±53 V loaded. For 4 Ω high-current setups, jump to 800 VA. Keep inrush under control with an NTC or soft-start relay [Elliott, 2022].

How does this design compare with an LME49830 + lateral MOSFET monoblock?

The LME49830 stage reaches 0.0007 % THD and needs only two output devices for 150 W. Thermal stability is superior because lateral MOSFETs have positive temp-co above 0 °C. Parts remain in production, but cost 30 % more and demand a separate Vgs tracking network [Rothacher, 2021].

Generated by the language model.