This page was last updated:  10/21/10
 

Posted below are some detailed Q&As relating to the design of the 300 Series of navigation instruments.

I notice that the new 302 manual refers to problems with some Prandtl/triple probes from the small static holes and potential pitot and static flow resistance mismatches. Can you be specific about any particular makes of probe to be avoided?

The Prandtl tube issue is perhaps not as serious as was implied in our User's Guide. Here is the background:

The 302 samples static pressure (altimeter) and dynamic pressure (pitot - static) 64 times per second. The Prandtl tube orifices together with the air volume enclosed in the tubing from the probe to the instruments constitute a low pass filter (equivalent to an electrical resistor/capacitor filter). If, for example, the time constant of the static port orifice and its tubing is different than the time constant for the pitot port, then rapid changes in dynamic pressure will not be measured correctly.

A mechanical airspeed indicator has very different internal volumes for its static port (the case -- large) and its pitot port (the bellows -- small. If one connects a prandtl tube to a mechanical airspeed indicator, the high resistance of the Prandtl tube orifices will cause different time constants for the two elements of dynamic pressure. This doesn't matter if the measuring instrument response time is very slow compared to the above mentioned time constants.

So far, I have alerted you to POSSIBLE Prandtl tube issues. Over the past several weeks, I've measured the flow resistance of several Prandtl tubes. I don't know the source for these tubes. I found that static and pitot flow resistances match reasonably well at low flows. In any event, it should be possible to "tune" the Prandtl tube by enlarging either the Pitot or Static orifices.

In general, I feel the quality of pressure measurement from a Prandtl tube is superior to that from fuselage statics, and would recommend incorporating it in any new glider.

Q. I thought Cambridge used to use flow sensing technology. As I read this, it seems that you have converted to altitude derivative with various enhancements. It sounds like you have special filters and filter logic. Can you explain what you are attempting to do?

A.  Raouf Ismail developed the first Cambridge Vario (CVS) in 1972 from a design he worked on as a student at Cambridge University in England. The CVS uses a thermistor-based flow sensor. This sensor technology was refined from 1980 to 1985. It has been unchanged since its 1986 application in the S-NAV.

Dave Ellis designed the first Cambridge pressure sensor vario (CPT) in 1979. As with all known pressure sensor varios, an analog resistor/capacitor differentiator circuit was used to take the derivative of the absolute pressure. This signal is not the same as the derivative of altitude (the two are related by a 5th order polynomial that describes the "standard" atmosphere). To altitude-compensate a pressure sensor vario, one must resort to an analog multiplier that changes the vario gain with altitude. In analog designs such as this, vario calibration depends on components with +/- 10% tolerance. Further, there is an awkward tradeoff between vario speed, noise coming from turbulence at the pneumatic ports, noise originating in the pressure sensor, and noise from the front-end amplifier. Traditional pressure sensor varios have 1 to 2 second time constants. In these designs there is no easy way to incorporate additional variables such as acceleration into the variometer computation. Perhaps more importantly, there is no way to modify the circuitry of production instruments as designers discover new ways to improve variometer performance.

It has been a long term Cambridge goal to sense pressure altitude directly with an Analog-to-Digital (A/D) converter having both resolution and speed adequate for variometer calculations to be done in software rather than analog hardware. We built a prototype direct digital variometer in 1986, but with A/D converters and microprocessors available at that time, performance was marginal and cost was prohibitive. Astounding progress on both fronts has been made in the last 15 years. The Cambridge 301 Direct Digital Variometer (DDV) uses high-resolution A/D converters for both pressure altitude and airspeed. Multi-stage digital signal processing yields pressure altitude with short-term, sea level altitude resolution of better than 2" (5cm) at 50 samples per second!  Based solely on the accuracy of the 301 altimeter sensor, we have finally realized our dream of a very fast, self-calibrating, altitude compensated variometer. Because filtering and computation are done in software, and because the new product has field-upgradeable flash program memory, we can deliver performance upgrades over the Internet.

Q. Is this an electronically compensated or a TE probe compensated system?

A. The 301 senses Dynamic pressure (Pitot - Static) and Static pressure with matched A/D converters and filters. It can be configured as either an electronic or a TE probe compensated variometer.  When configured for electronic TE, the absolute pressure sensor is connected to a static port.  When configured to use the TE probe, the absolute pressure sensor is connected to the TE probe.

Q. You use a two-axis accelerometer as part of the vario circuit? What is that doing? Are you integrating to get the lift component on one axis and using the other orthogonal axis to do the TE correction? Of course you take vectors as the glider could be pitching. What puzzles me is that the web page says the accelerometers are used to supplement [data from] the pressure sensors used to get altimeter and airspeed. Surely you mean for the vario which derives its info from altitude, airspeed, and TE probe.

A.  The pilot's "backside" is sensitive to the "feel' of the air. Good pilots make use of this sensitive, high-bandwidth information channel to aid their flying. Our goal is to make the audio variometer time-correlated with the pilot's "backside". We feel this will make it easier for pilots to understand the signal coming from their butts. However, unless we address problems such as sensitivity to horizontal gusts, we will drive pilots crazy with manic audio noise.

Vertical and fore/aft acceleration are sensed over a +/- 2 g range with ~ 10 bit resolution. The data rate is ~ 50 samples per second for each axis. This means accelerations can be combined with pneumatic signals at an early stage in digital signal processing. Here is an example of how a vertical axis accelerometer can be used to improve TE variometer performance:

Do a 1.5-g pull-up into a thermal. The TE probe provides first-order compensation for the glider's Total Energy. However, the 1.5-g pull-up requires that the wings produce both extra lift and induced drag. The TE probe cannot account for the energy loss due to extra induced drag. Given a perfectly compensated TE probe, you will see a real negative vario deflection proportional to wing loading. For a given mass, vertical acceleration is proportional to wing loading, so it can provide a second-order correction to the variometer reading.  This enhancement is described in US patent # 5,175,540. It should make thermal entry and centering easier.

Here is an example of how a fore/aft-axis accelerometer MAY help reduce a TE vario's sensitivity to horizontal air mass gusts:

An interesting experiment is to switch the vario input from the TE probe to a static port while climbing in a thermal. If you keep a steady hand on the stick, you will find the audio is less jittery when connected to the static port. This is because the TE probe is equally sensitive to changes in dynamic and static pressure.

Gliders are fun because they are very slippery along the fore/aft axis. In smooth air, fore/aft acceleration is just the first derivative of airspeed. A horizontal gust causes a rapid change in measured airspeed without a corresponding change in fore/aft acceleration. It SHOULD be possible to utilize the fore/aft acceleration to discriminate between actual changes in the glider's Total Energy (referenced to ground coordinates), and apparent energy changes due to the short-term gusts that affect dynamic pressure.

At present we don't have enough data to design a variometer algorithm that fully utilizes acceleration as well as pneumatic signals. We will use the 301 and 304 with special software to record high bandwidth sensor outputs during normal glider flight in a variety of conditions. Based on analysis of this data, we will attempt to improve on the traditional TE Variometer.

We are very encouraged by the variometer performance we've already obtained with fairly simple filter algorithms. Because we have not yet incorporated acceleration readings in the variometer design, we cannot predict the time scale for future software performance upgrades.

Q. Maybe it is a non-issue, but your dial shows 10 knots through a 180 degree rotation vs the old "standard" 140 or so degrees. Gut feel is that people used to the old style might find this a little confusing. The instrument scale is shown as +/- 10 knots, but does it have a dual range?

A.  Our primary vario software designer has flown the 180 degree vario dial without noticing the transition.  However, one could say that he is biased in favor of our instrument design!

We recognized vario pointer angle as a potential problem and weighed the alternatives. A 301 design goal is elimination of the range switch. The 301 variometer pointer is driven with a stepper motor. The pointer rotates through 360 degrees, so we can get rid of the range switch.  For lift greater than 10 knots, the pointer goes beyond 180 degrees. When it points straight down you either going up at 15 knots or down at 5 knots. The audio pitch for these two states is VERY different. We think pilots will have no problem distinguishing super lift from moderate sink or vice-versa.

Q. A numeric readout for averager? There is a much better way to do that.

A.  The yaw string is arguably the most important gliding instrument. But we can't make much money selling yaw strings! The vario is certainly the next most important gliding instrument. We've been building varios for more than 20 years -- long enough to have evolved simple guidelines for communicating lift information to the pilot. You may not agree with them, but if you understand these guidelines, you'll know where the 301 design came from.

There are 4 pathways for communicating lift to the pilot:

a. The pilot's "backside" is exquisitely sensitive to CHANGES in vertical acceleration. To “feel” this, one must be in a state of intense relaxation. Because most of us become tense when we get very low, it is very easy to lose this “feeling” when it is most needed.

b. The audio variometer is a continuously available, high-bandwidth channel. The ear can easily detect CHANGES in lift. Unless you are blessed (or cursed) with perfect pitch or rhythm sense, the audio variometer is not calibrated.  This channel is not disturbed when your body becomes tense.

c. The vario pointer is calibrated, but if you watch it continuously, your eyes won't available for looking at the next cloud or avoiding mid-air collisions. The 301 vario pointer is as big and bright as we can make it.
A quick glance or even one's peripheral vision calibrates what you hear. There is almost no value in remembering a numerical variometer reading because it changes so quickly. It is better to remember the pointer angle. 301 dial markings are not very bold because we prefer to have pilots sense the angle of the moving pointer.

d. Average lift changes slowly enough that remembering the numerical value is both practical and useful. We display the average as n.n kts. to make this easier. Imagine trying to remember n.nn!  A useful guide is to note the reading once per circle in climb. The difference in reading helps one make centering decisions. For mentally challenged pilots such as this author, it is sometimes hard to remember the previous averager reading. For this reason the 301 adds averager trend "chevrons" similar to those found on the old Peschges LCD vario. A single glance at the screen yields both the number and a sense that the number is increasing, decreasing, or stable.

The averager number is what pilots remember thermal-to-thermal; it is also used as a guide to setting the MacCready value for the day.  Many pilots form an opinion of lift quality for the day from the averager. They often leave thermals when lift falls below the numerical averager threshold of “good” for the day. These are reasons why we feel that numbers are the best way to display average lift.

Cambridge gliding instruments have followed these guidelines for many years.  The 300 series makes no radical changes.  Instead, it makes subtle improvements in 3 of the 4 ways pilots understand lift: Audio pitch and interrupt rate, analog vario pointer angle, and digital average.

Q. I used to think a fast vario was what I wanted. I then began to realize that I probably liked a slower, smoother variation. Does the 301 allow the pilot some control over the vario time constant? The only way I can see you making an immediate lift readout is with the accelerometers. Then you go on to talk about the Schuemann research and how 1.2-1.5 seconds is optimal. So, is it an immediate readout or a delayed readout?

A. We have learned that pilots react very subjectively to variometer audio behavior. We know one world class pilot who clings to his 30 year old audio despite the more pleasant sounds emanating from his whizzy new glide computer. For most pilots, the transition from an audio vario with a 1.5 second time constant to one with a 0.5 second time constant will be nerve wracking to say the least!

The benefit of a really fast vario is in the temporal correlation between what you feel with your butt and what you hear with your ears. This starts to "click" when the time constant approaches 0. 5 seconds. To help pilots make the transition from their old audio to the 301, we provide time constant adjustment over a wide range using the front panel control knob.

Out of curiosity who were the pilots who flight tested the new vario?

Prototype sensor platforms were built in 1999. They were implemented in the L-NAV case, so the DDV could be compared with the traditional Cambridge flow meter variometer. First test flights in September, 1999 were by Dave Ellis, the owner of Cambridge. A prototype was flown in Australia over the winter of 1999/200. Dave again flew a prototype DDV in a Duo-Discus with Makoto Ichikawa, our Japanese agent. Mak is a full-time glider pilot, and member of the Japan glider team who spends the winter in Australia teaching cross-country gliding to Japanese pilots. He is a disciple of Ingo Renner. Ingo prefers to fly without audio, and teaches pilots to rely, instead, on the "feel" of the air. "Mak" Ichikawa confirmed our suspicions that the fast DDV audio is a significant achievement.

Dick Butler (ASW-22 and ASW-27) flew during 2000 season with a prototype DDV. Dick worked on varios with Wil Schuemann in the 70's. Dick is fully aware of the technical issues, and is enthusiastic about DDV performance.

Chip Garner is the principal designer of the Cambridge DDV filter algorithms. His test gliders are a Grob 103, Discus a, and DG-800s.

Q. My preference would be to have the vario needle (analog display) of the 301 switch to speed-to-fly directive (push-pull) when in cruise mode. An added nice feature would be to have the LCD display of averager switch to relative vario (super-netto) when in cruise mode. What are your thoughts on this?

Chip Garner's comments are:

The 300 series speed to fly will work very similarly to the L-NAV. We plan to always have the needle show the variometer, with a push or pull arrow and audio tones for speed to fly in cruise. The L-NAV, has multiple speed bars to indicate how much to speed up or slow down, but we have found that most pilots ignore them and rely primarily on the audio.

Following speed to fly with the needle can lead to confusion, and doesn't really add information to the audio. The 300 will be configurable with pilot's preferences via a PC, and we will add this to the preferences if we get several requests for it.

Initially, the instrument will work as follows:

Above best L/D, the vario becomes relative. This is a smooth transition. The averager is always an averager, and is this showing relative average at high speeds. On switching from cruise to climb, the averager is reset to the climb rate at the time of the switch and begins averaging again from there. We have found that this works very well, giving a good average after less than half a turn when you actually turn into a thermal without missing.

Any time the climb rate is positive, the audio switches to vario mode. This is because the speed to fly requires more damping (eg 4 seconds) and you don't want to fly past a thermal before you find out about it. The time constant difference is another reason we like keeping the needle on vario all the time. Both the averager and the speed to fly are too slow to help with the very crucial decision of whether or not to turn in a thermal.

Q. The recent "Themi" thermal director apparently uses GPS information to figure out how the pilot should turn in the thermal. Does the 300 series "paint" a picture of the trajectory of the glider along with lift/sink readings? Can the scale of the NAV portion be reduced to the point where one can actually see the circling in the thermal? That would require drift correction as well.

A.  The Themi computes distance and bearing to the center of an assumed round thermal using ground speed, track, and altitude. Estimated wind drift is included in the computation. Two "HUD-like" LED’s on the instrument panel top serve as the user interface. LED blink patterns coach the pilot to turn more or less tightly so the glider is led towards the thermal center. I've flown several hours with the Themi. Both the concept and the implementation show promise. The 302 design extends naturally into functions that resemble the Themi. We have no current plans for this work.

A "God's-eye" graphic display of circling flight is required to "paint" the picture you describe. The WinPilot and German Cenfis glide computers both offer this display. The Cambridge 304 Pocket-NAV certainly has the capability for displaying climb information in this manner. Despite interest shown by some potential customers, we have decided NOT to provide this function. The reason is safety. Such displays encourage staring at the instrument rather than the airspace. We feel strongly that this is dangerous. OSTIV, the Scientific Committee of the IGC, agrees with us in a 1999 special safety report.

The 302  transmits both the NMEA RMC and GGA sentences. This makes it compatible with Winpilot.