Heater Delay Timer

This article describes a flexible timer for delaying PTT operation of a valve amplifier until sufficient time for the cathode to reach operating temperature.

Design criteria

The timer sits between the low voltage positive DC source used to control TR relays and the relays themselves. Note that if the same DC source is used for other purposes (eg bias, meter lamps, ALC delay, bias etc), the supply to those functions must be tapped off before the timer.

Key design criteria were:

  1. operation from +12 to +20V DC supply;
  2. load current capacity to 200mA;
  3. delayed switching of input supply to output based on timer operation;
  4. flexible, adjustable delay interval;
  5. reliable;
  6. simple in construction and adjustment;
  7. low power requirements; and
  8. capable of retrofit to existing amplifiers to ensure adequate cathode warm up time at initial power on.



Fig 1: Circuit diagram

The heart of the timer is a ATTiny25 microcontroller (MCU). The MCU starts a timer when it is initialised at power on, and after a predetermined delay (fetched from MCU EEPROM), asserts PORTB4 which causes the BC337 to conduct, which in turn causes the BD140 to conduct. The transistors are both saturated, so they operate as switches.

Fig 1 shows the circuit diagram.

As stated, the required delay is stored in EEPROM. The value is configured by closing the CFG link. The link is then opened for the desired time and reclosed. The period is measured, rounded and written to EEPROM. When the link is then opened, the circuit will restart with the newly configure delay time. Link L1 must be left open when configuration is complete, the timer will not operate normally with L1 closed.

Link L2 can be used to select between two timer ranges:

The timer can easily be programmed (and reprogrammed) in this way for say, 10s for 811A, 60s for 6146B, 180s for 4CX250B.

The MCU is used in internal RC oscillator mode. Maximum error is about 10%, but usually much less. To some extent, the piece to piece variation is compensated by the method of measuring and storing the required delay time. The MCU goes in to SLEEP mode when the time expires, meaning that the oscillator ceases to run, reducing the risk of radiation of radio frequency energy.

Fig 2: Component side of Verobard prototype


The MCU was socketed on the prototype to permit easy replacement of the microcontroller. Reliability is enhanced by soldering the chip directly to the board. The 6 pin header is for In-System-Programming, and was added because this is a prototype.


Fig 3: Copper side of Veroboard prototype

Note that tracks are cut between holes in two groups, one under the ISP header and the other between the ISP header and MCU. A construction without the ISP header would not require cutting tracks other than at holes. The tracks to the right of the top left mounting hole need to be cut to allow the use of a metallic mounting pillar.

Fig 4: Component side of Verobard prototype - layout #2

Above is a slightly different layout developed in a presentation tool. C4 is not shown on the circuit diagram, it is 0.001µF between collector of Q2 and ground.

Fig 3: Copper side of Veroboard prototype - layout #2

And above the copper side of layout #2.

Fig 6: Installation in AL-80B.

Above is an installation in the author's AL80B. The HDT is mounted on a tinplate bracked sucured to the mounting screws for the GOP-100 board. A green LED was installed above the power switch and connected to the HDT output via a 1k resistor. The green LED lights when the timer has expired and relay voltage is available to operate the amplifier. Again, this is a prototype board and the ISP socket (at the left) is not necessary for typical applications, so a smaller board is possible.

A simpler implementation is to build the basic timer with three flying leads, encapsulate it with heat shrink once adusted and use double sided foam tape to secure it to a convenient surface.

Frequently asked questions (FAQ)

Can the circuit be used at higher control circuit voltages?

The circuit was designed for nominal +12V control circuits, and will operate at up to +20V safely. To use it at higher voltages, the two transistors and their base resistors may need to be revised. For operation from +20V to +40V DC, the existing transistors are ok, the BC337 collector resistor and the 820Ω resistor should be doubled, the 470 feeding the zener needs to be increased to 1k to 2k, and the 1k resistor in the collector of the BC337 should be increased to 3.3k.

Variations on output switching

The output switching circuit above is effective, it has low voltage drop and good voltage rating. Other options include Solid State Relays, though those with similar ratings are not DIP size packages. DIP reed relays such as the CP Clare PRMA 1A05 could be driven directly by the MCU, don't forget a quench diode.

Why does my timer never expire?

Is CFG closed? If the timer is started with CFG closed, it puts the MCU into configuration mode and will not expire until the configuration process is completed. If the configuration has been performed, remove power, open CFG, and reapply power for normal operation.

Is the timer compatible with the Yaesu FL-2100Z?

Whilst it has not been tested, the timer should be compatible with the FL-2100Z. Use the timer to control the DC supply to the coil of Relay RL1 only.

Isn't an MCU overkill, won't an NE555 work fine?

Though you will see claims that NE555s work for times up to hours, there are some practical issues with the NE555 that limits reliable repeatable times to less than perhaps 100s. For long intervals, the product of timing C and charing R must be large, and leakage of C must be much smaller than the charging current.

 Therein lies the problem, for a target time of say 180s from a 12V rail, lets say you choose a 1MΩ charging resistor, the current in the resistor at trigger voltage (2/3 supply) is 4µA. So, the C required will be 160µF, it is unlikely that the leakage current of an ordinary 16V Tantalum cap at 8V will be reliably less than 4µA, and it needs to be much less than 4µA for reliable operation. Sure, you could decrease R to swamp the leakage, but then you need to increase C and you are right back where you started. (Check the price of large Tantalum caps, they don't come cheap!)

Lots of hams have published timer circuits using inexpensive electrolytics for the timer C. Not only is their leakage a bigger problem, but their tolerance is poorer, temperature dependence worse, and history effects degrade repeatability.

Isn't the MCU internal RC based oscillator temperature sensitive?

Yes, the datasheet show the temperature coefficient of clock frequency to be 0.05%/°C, or about 2% from 20 to 60°C. This is quite adequate for the purpose, and far better than achieved with a single stage NE555 timer for long intervals.

Sourcing parts

All the parts should be easily obtained, or easily substituted.

The hex code for the MCU licenced for personal use ONLY can be downloaded here.

Programmed MCU chips are available from the author for US$6.50 each + US$4 P+P within Australia, New Zealand, UK, and USA, postage to other destinations will depend on the destination. This is NOT a kit of parts, just a programmed PIC chip. Preferred method of payment is Paypal, and Paypal is the only option accepted for international payers. Orders are usually despatched within 48 hours, but may sometimes be delayed by up to two weeks. 

© Copyright: Owen Duffy 1995, 2017. All rights reserved. Disclaimer.