Forums :: NEW! Web Resources :: Features :: Photo Gallery :: Vintage Radio Shows :: Archives
Support This Site: Contributors :: Advertise


It is currently Oct Thu 28, 2021 5:34 am


All times are UTC [ DST ]





Post New Topic Post Reply  [ 24 posts ]  Go to page Previous  1, 2
Author Message
 Post subject: Re: Choke hum on LM386 Transmitter
PostPosted: Aug Thu 19, 2021 2:19 pm 
Member
User avatar

Joined: Jun Sun 22, 2008 2:03 am
Posts: 1062
Location: Burke, VA 22015
I'm glad to see it working. However there are a few technical issues with your design.

Since the audio is fed to the collector (or drain) through a large choke, the modulator is a current-fed type. Its output impedance is high. The antenna load presents a lower impedance. Therefore a high-to-low impedance matching network is necessary. Usually it's a pi or a shunt first CL network. A series first LC lowpass network, as you currently use, is normally used for low-to-high impedance matching. Without a capacitor shunt from the collector to the grounded emitter, the choke becomes part of the matching network when the transistor is off. A 2mH choke resonates with a 25pF capacitor at 1 MHz. The tuning inductor will lower the capacitance further, making it impossible to match to an antenna of higher capacitance. Thus a shunt capacitor will decouple the choke from the tuning network while also serving as a snubber to lower the flyback voltage.

Macrohenry wrote:
The output of the frequency synthesizer is a curved triangle wave

That's likely what saves the transistor from being destroyed by the very high flyback voltage from the choke when its current is abruptly disrupted. Since the transistor is soft driven by a triangular voltage, the choke's flyback voltage is much lower. As Hugo explained earlier, the capacitive coupling without a discharge resistor at the base no longer passes the AC drive signal in the steady state. Therefore the transistor is weakly driven and the flyback voltage is kept below the transistor's destruction voltage.

A BJT is a current-controlled current source. It's better driven by a current through a series resistor from a voltage source. An anti-parallel diode across B & E is a must to prevent the base breakdown the driven voltage is too high. This diode is especially important if you decide to stay with the simple capacitive coupling. With 12V logic, the base will see a negative voltage of 11.3V and will likely zener. A B-E resistor maintains the discharge path when both the base and diode are off.

All of these that happened to protect the BJT by chance do not matter if the choke happens to resonate with the LC network at the driven frequency. The choke's voltage will likely exceed the transistor C-E breakdown voltage. A higher voltage device such as a tube or power MOSFET can tolerate it better than a low-voltage BJT.

A bypass capacitor at the audio feed input is absolutely necessary. It supplies AC current to the modulator and prevents the RF signal from getting back to the audio PA. A small series resistor isolates the capacitive loading from the audio PA output for stability.

Putting all of this together, below is a circuit example that is relatively simple and works reasonably well (at least in simulation). RF harmonic rejection is better than 50dB at 1MHz. Audio THD is below 3% at 1kHz. The DC input power is less than 22mW at 6V idle and 38mW at 90% modulation. The collector voltage is below the breakdown value even during start up. The RF drive current is well within the capability of a CD family driver. The C-E shunt capacitor is about ten times the antenna equivalent capacitance to minimize detuning when the transistor is off. The tuning capacitor, which is not needed when the tuning is done inductively, should be kept to a mininum. The transistor operates within its safe ratings even when the antenna is detached, the matching network is completely mistuned or the transistor is hard driven by an RF square wave with fast edges.

Hope this helps and it's your choice to decide whether it's too complicated. I respect your wish to keep this thing simple. However please note that there are design elements that cannot be omitted without compromising reliability.
Attachment:
ClassC_E_AM_1.png
ClassC_E_AM_1.png [ 15.47 KiB | Viewed 318 times ]

_________________
Binh


Top
 Profile  
 
 Post subject: Re: Choke hum on LM386 Transmitter
PostPosted: Aug Thu 19, 2021 5:38 pm 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 4963
Location: Texas
Excellent, Binh, thank you for the engineering assistance!

I'm very curious how each of these changes affects things. The good news is there's still room on the board to add these inexpensive components. A previously unstated design goal is to use this exact board size because the PCB manufacturer offers a discounted rate for it.

It will take a couple days to get to this. It all makes sense and thank you for your contribution.

In the meantime, I do have one question. The B-E resistor is 1K, a far cry from the 100K previously mentioned to keep a FET gate from floating. I've been designing the board so different FETS (both gds and dgs) and BJTs can be used. If a FET is used, can that 1K be kept, or would it be better to specify a 100K if a FET is used?


Top
 Profile  
 
 Post subject: Re: Choke hum on LM386 Transmitter
PostPosted: Aug Thu 19, 2021 7:06 pm 
Member
User avatar

Joined: Jun Sun 22, 2008 2:03 am
Posts: 1062
Location: Burke, VA 22015
Macrohenry wrote:
The B-E resistor is 1K, a far cry from the 100K previously mentioned to keep a FET gate from floating. I've been designing the board so different FETS (both gds and dgs) and BJTs can be used. If a FET is used, can that 1K be kept, or would it be better to specify a 100K if a FET is used?

Your RF synthesizer has a weak output driver. It can sink or source a few mA (2 - 4mA). When driving a MOSFET, that current is not high enough to discharge the gate-source capacitance Cgs within the half cycle of the driving frequency. After a while from start-up, Cgs charges up to the driving voltage and prevents the gate from turning off. Therefore the grid leak resistor (R2) has to be low enough to discharge Cgs on time while not loading down the driver voltage below the gate turn-on threshold. A 10 kOhm resistor for the grid leak will work for a 2N7002 MOSFET as well as a BJT. The grid stop resistor (R1) needs to be reduced to 100 Ohm (exact value not important). Note that the restriction on R2 does not apply with a strong gate driver capable of charging/discharging a large gate capacitance typical of a power MOSFET (hence the previously mentioned 100K resistor which is only there for gate protection).

_________________
Binh


Top
 Profile  
 
 Post subject: Re: Choke hum on LM386 Transmitter
PostPosted: Aug Tue 24, 2021 3:21 am 
Member
User avatar

Joined: Jan Thu 01, 1970 1:00 am
Posts: 4963
Location: Texas
Excellent explanation, thanks.

I've ordered some 2N7000, as I see them in 386 circuits all over. Meanwhile I've tried bipolar and FET. Here are my best performing in each category and how the suggested changes affected operation.

MPSA18 bipolar

2K resistor added in series with base drops output slightly, negligibly for our purposes.
1K resistor added from base to ground very slightly increases output and provides a cleaner, more focused trace on the scope.
1N4148 from base to ground flatlines the output. It acts like a direct short. Curiously, forward biased there's still oscillation. just not enough to achieve resonance.
10 ohm resistor from 386 to choke slightly decreases output from 120 mW to 91 mW, It provides a handy way to measure collector current to ensure there's no more than 100 mW into the final.
.1 uF between the 10 ohm and choke doubles the entire current draw and heats up the 386 where I can smell it.

IRF520
100K resistor added from gate to ground has no visible effect on scope trace. There will be a place for it on the board for those who want to use a FET.
1N4148 from base to ground slightly changes tuning and output, negligible. Since it adds protection with little additional expense, I'll look for a place on the board for it.
10 ohm resistor from 386 to choke slightly decreases output from 125mW to 98 mW, It provides a handy way to measure collector current to ensure there's no more than 100 mW into the final.
.1 uF between the 10 ohm and choke doubles the entire current draw and heats up the 386 where I can smell it.

Thank you, Acornvalve and Binh for your concise explanations and suggestions to improve this circuit. An LED tuning circuit will be included so I'm running out of space on the board, but I should be able to add your improvements that keep it running.


Top
 Profile  
 
Post New Topic Post Reply  [ 24 posts ]  Go to page Previous  1, 2

All times are UTC [ DST ]


Who is online

Users browsing this forum: No registered users and 8 guests



Search for:
Jump to:  


































Privacy Policy :: Powered by phpBB